{"generated":"2026-10-18T22:01:20.992600+00:00","version":"0.1.0","excludedRuns":6,"datasets":[{"id":"opinionsqa","label":"OpinionsQA","population":"US adults (Pew American Trends Panel)","core":true,"questions":684,"random":0.762852,"majority":0.704504,"majorityIdx":-24.7,"ceiling":0.999518,"ceilingAssumed":false,"ceilingQuality":"high"},{"id":"subpop","label":"SubPOP","population":"22 US subpopulations","core":true,"questions":200,"random":0.757397,"majority":0.672732,"majorityIdx":-35.6,"ceiling":0.995346,"ceilingAssumed":false,"ceilingQuality":"medium"},{"id":"gss","label":"GSS","population":"US adults (General Social Survey)","core":true,"questions":75,"random":0.73538,"majority":0.721147,"majorityIdx":-5.4,"ceiling":0.998784,"ceilingAssumed":false,"ceilingQuality":"high"},{"id":"globalopinionqa","label":"GlobalOpinionQA","population":"Cross-national (Pew Global Attitudes, WVS)","core":true,"questions":100,"random":0.774017,"majority":0.68956,"majorityIdx":-37.8,"ceiling":0.997219,"ceilingAssumed":false,"ceilingQuality":"medium"},{"id":"ntia","label":"NTIA","population":"US adults (NTIA Internet Use Survey)","core":false,"questions":79,"random":0.788153,"majority":0.664986,"majorityIdx":-58.1,"ceiling":0.999973,"ceilingAssumed":false,"ceilingQuality":"high"}],"systems":[{"key":"ensemble|complementary-4|default|current|||","base":"Ensemble: Qwen 3.8 Flash (Althing), Sonnet 4 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","name":"Ensemble of Qwen 3.8 Flash (persona), Sonnet 4 (persona), DeepSeek V4.1 Flash (raw), Gemini 2.5 Flash-Lite (raw)","short":"Qwen + Sonnet + DeepSeek + Gemini","baseId":"ensemble-complementary-4","members":[{"id":"qwen/qwen3.8-flash","method":"althing"},{"id":"anthropic/claude-sonnet-4","method":"althing"},{"id":"deepseek/deepseek-v4.1-flash","method":"raw"},{"id":"google/gemini-2.5-flash-lite","method":"raw"}],"lab":"DataViking","method":"ensemble","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.830727,"idx":25.4,"lo":13.3,"hi":37.4,"pDist":0.795177,"pRank":0.717688,"pRefuse":0.979316,"n":100,"runs":1,"samplesPerQuestion":null,"cost100":null,"latencyP50":null,"badge":"verified","holdoutGap":0.047824,"holdoutLimit":0.1066,"topicIdx":{"politics":14.8,"economy":71,"technology":36.3,"science-health":57.2,"identity":-2,"society":75.3,"global":26.1},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"ensemble--complementary-4--tdefault--tplcurrent--2e4cb430"},"gss":{"sps":0.863134,"idx":48.5,"lo":40,"hi":56.8,"pDist":0.817346,"pRank":0.800115,"pRefuse":0.971939,"n":75,"runs":1,"samplesPerQuestion":null,"cost100":null,"latencyP50":null,"badge":"verified","holdoutGap":0.040697,"holdoutLimit":0.129099,"topicIdx":{"politics":46.4,"economy":39.4,"technology":62,"science-health":74.1,"identity":34.6,"society":61.9,"global":11.6},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"ensemble--complementary-4--tdefault--tplcurrent--fb17ebf4"},"opinionsqa":{"sps":0.912853,"idx":63.4,"lo":57.9,"hi":68.8,"pDist":0.87428,"pRank":0.870007,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":null,"cost100":null,"latencyP50":null,"badge":"verified","holdoutGap":0.004523,"holdoutLimit":0.104257,"topicIdx":{"politics":32.1,"economy":30.7,"technology":61.5,"science-health":66.2,"identity":59.4,"society":73.9},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"ensemble--complementary-4--tdefault--tplcurrent--1c6ab6e1"},"subpop":{"sps":0.864985,"idx":45.2,"lo":35.3,"hi":53.4,"pDist":0.801653,"pRank":0.813203,"pRefuse":0.9801,"n":100,"runs":1,"samplesPerQuestion":null,"cost100":null,"latencyP50":null,"badge":"verified","holdoutGap":0.037781,"holdoutLimit":0.094491,"topicIdx":{"politics":34.1,"economy":65.9,"technology":41.9,"science-health":72.2,"identity":47.5,"society":47.7,"global":62.4},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"ensemble--complementary-4--tdefault--tplcurrent--d17dc795"}},"coverage":4,"index":45.6,"indexLo":41.1,"indexHi":50.1,"cost100":null,"latencyP50":null,"flagged":0,"runs":4},{"key":"ensemble|complementary-top|default|current|||","base":"Ensemble: Haiku 5.5 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","name":"Ensemble of Haiku 5.5 (persona), DeepSeek V4.1 Flash (raw), Gemini 2.5 Flash-Lite (raw)","short":"Haiku + DeepSeek + Gemini","baseId":"ensemble-complementary-top","members":[{"id":"anthropic/claude-haiku-5-5","method":"althing"},{"id":"deepseek/deepseek-v4.1-flash","method":"raw"},{"id":"google/gemini-2.5-flash-lite","method":"raw"}],"lab":"DataViking","method":"ensemble","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.825182,"idx":22.9,"lo":10.4,"hi":35,"pDist":0.794438,"pRank":0.70192,"pRefuse":0.979187,"n":100,"runs":1,"samplesPerQuestion":null,"cost100":null,"latencyP50":null,"badge":"verified","holdoutGap":0.031741,"holdoutLimit":0.1066,"topicIdx":{"politics":17.6,"economy":40.1,"technology":31.6,"science-health":58.8,"identity":1.3,"society":56.7,"global":22.7},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"ensemble--complementary-top--tdefault--tplcurrent--ee30a5f9"},"gss":{"sps":0.851459,"idx":44.1,"lo":35.9,"hi":51.9,"pDist":0.799069,"pRank":0.783968,"pRefuse":0.97134,"n":75,"runs":1,"samplesPerQuestion":null,"cost100":null,"latencyP50":null,"badge":"verified","holdoutGap":0.040717,"holdoutLimit":0.129099,"topicIdx":{"politics":46.9,"economy":23.2,"technology":56.1,"science-health":61.7,"identity":31.4,"society":53.1,"global":27.3},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"ensemble--complementary-top--tdefault--tplcurrent--cab9400c"},"opinionsqa":{"sps":0.905567,"idx":60.3,"lo":54.1,"hi":66.1,"pDist":0.870101,"pRank":0.852668,"pRefuse":0.993933,"n":100,"runs":1,"samplesPerQuestion":null,"cost100":null,"latencyP50":null,"badge":"verified","holdoutGap":0.000368,"holdoutLimit":0.104257,"topicIdx":{"politics":69.8,"economy":42.2,"technology":55.2,"science-health":63.2,"identity":57,"society":72.2},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"ensemble--complementary-top--tdefault--tplcurrent--93ab1723"},"subpop":{"sps":0.856103,"idx":41.5,"lo":32.8,"hi":48.8,"pDist":0.791072,"pRank":0.799167,"pRefuse":0.978068,"n":100,"runs":1,"samplesPerQuestion":null,"cost100":null,"latencyP50":null,"badge":"verified","holdoutGap":0.036671,"holdoutLimit":0.094491,"topicIdx":{"politics":29.9,"economy":61.2,"technology":39.3,"science-health":74.7,"identity":40,"society":45.4,"global":57.5},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"ensemble--complementary-top--tdefault--tplcurrent--dfa9f8ed"}},"coverage":4,"index":42.2,"indexLo":37.8,"indexHi":46.6,"cost100":null,"latencyP50":null,"flagged":0,"runs":4},{"key":"ensemble|flash3-raw|default|current|||","base":"Ensemble: Qwen 3.8 Flash, DeepSeek V4.1 Flash, Gemini 3.5 Flash-Lite (raw)","name":"Ensemble of Qwen 3.8 Flash, DeepSeek V4.1 Flash, Gemini 3.5 Flash-Lite (all raw)","short":"Qwen + DeepSeek + Gemini, raw","baseId":"ensemble-flash3-raw","members":[{"id":"qwen/qwen3.8-flash","method":"raw"},{"id":"deepseek/deepseek-v4.1-flash","method":"raw"},{"id":"google/gemini-3.5-flash-lite","method":"raw"}],"lab":"DataViking","method":"ensemble","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.817169,"idx":19.3,"lo":6.9,"hi":31,"pDist":0.779284,"pRank":0.698618,"pRefuse":0.973604,"n":100,"runs":1,"samplesPerQuestion":null,"cost100":null,"latencyP50":null,"badge":"verified","holdoutGap":0.0083,"holdoutLimit":0.1066,"topicIdx":{"politics":13.9,"economy":43.2,"technology":37.6,"science-health":43.5,"identity":-3.6,"society":56.7,"global":18.5},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"ensemble--flash3-raw--tdefault--tplcurrent--2f332ba9"},"gss":{"sps":0.844613,"idx":41.5,"lo":33.8,"hi":48.3,"pDist":0.777088,"pRank":0.784701,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":null,"cost100":null,"latencyP50":null,"badge":"verified","holdoutGap":0.024824,"holdoutLimit":0.129099,"topicIdx":{"politics":47,"economy":31.9,"technology":43.6,"science-health":56.5,"identity":32.9,"society":39.7,"global":25.4},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"ensemble--flash3-raw--tdefault--tplcurrent--e43ae6b8"},"opinionsqa":{"sps":0.890868,"idx":54.1,"lo":48.2,"hi":59.5,"pDist":0.842973,"pRank":0.835358,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":null,"cost100":null,"latencyP50":null,"badge":"verified","holdoutGap":0.005271,"holdoutLimit":0.104257,"topicIdx":{"politics":70.5,"economy":40.9,"technology":49,"science-health":59.8,"identity":60.3,"society":58.3},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"ensemble--flash3-raw--tdefault--tplcurrent--3a43b801"},"subpop":{"sps":0.859209,"idx":42.8,"lo":35.4,"hi":50.1,"pDist":0.793484,"pRank":0.804102,"pRefuse":0.980042,"n":100,"runs":1,"samplesPerQuestion":null,"cost100":null,"latencyP50":null,"badge":"verified","holdoutGap":0.041311,"holdoutLimit":0.094491,"topicIdx":{"politics":49.1,"economy":62.1,"technology":31.4,"science-health":54.4,"identity":34.4,"society":54,"global":48.7},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"ensemble--flash3-raw--tdefault--tplcurrent--a416f83d"}},"coverage":4,"index":39.4,"indexLo":35.2,"indexHi":43.6,"cost100":null,"latencyP50":null,"flagged":0,"runs":4},{"key":"ensemble|complementary-mid|default|current|||","base":"Ensemble: Haiku 5.5 (Althing), Llama 3.3 70B, Gemini 2.5 Flash-Lite","name":"Ensemble of Haiku 5.5 (persona), Llama 3.3 70B (raw), Gemini 2.5 Flash-Lite (raw)","short":"Haiku + Llama + Gemini","baseId":"ensemble-complementary-mid","members":[{"id":"anthropic/claude-haiku-5-5","method":"althing"},{"id":"meta-llama/llama-3.3-70b-instruct","method":"raw"},{"id":"google/gemini-2.5-flash-lite","method":"raw"}],"lab":"DataViking","method":"ensemble","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.818198,"idx":19.8,"lo":7,"hi":32.4,"pDist":0.777522,"pRank":0.697885,"pRefuse":0.979187,"n":100,"runs":1,"samplesPerQuestion":null,"cost100":3.323261,"latencyP50":null,"badge":"verified","holdoutGap":0.041413,"holdoutLimit":0.1066,"topicIdx":{"politics":17.9,"economy":36.9,"technology":25.5,"science-health":48.9,"identity":-6.6,"society":56.7,"global":19.3},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"ensemble--complementary-mid--tdefault--tplcurrent--6fdfc610"},"gss":{"sps":0.847516,"idx":42.6,"lo":33.8,"hi":50.1,"pDist":0.784758,"pRank":0.786451,"pRefuse":0.97134,"n":75,"runs":1,"samplesPerQuestion":null,"cost100":2.579913,"latencyP50":null,"badge":"verified","holdoutGap":0.037585,"holdoutLimit":0.129099,"topicIdx":{"politics":50.3,"economy":19.1,"technology":54.6,"science-health":58.4,"identity":25.8,"society":51,"global":24},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"ensemble--complementary-mid--tdefault--tplcurrent--faac0e30"},"opinionsqa":{"sps":0.897395,"idx":56.8,"lo":49.3,"hi":63.3,"pDist":0.859192,"pRank":0.83906,"pRefuse":0.993933,"n":100,"runs":1,"samplesPerQuestion":null,"cost100":1.899411,"latencyP50":null,"badge":"verified","holdoutGap":0.035711,"holdoutLimit":0.104257,"topicIdx":{"politics":47.1,"economy":34.3,"technology":49.1,"science-health":61.7,"identity":48.9,"society":76.6},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"ensemble--complementary-mid--tdefault--tplcurrent--a91d1824"},"subpop":{"sps":0.845827,"idx":37.2,"lo":28.8,"hi":44.3,"pDist":0.783219,"pRank":0.776192,"pRefuse":0.978068,"n":100,"runs":1,"samplesPerQuestion":null,"cost100":2.532115,"latencyP50":null,"badge":"verified","holdoutGap":0.023043,"holdoutLimit":0.094491,"topicIdx":{"politics":24.8,"economy":72,"technology":34,"science-health":63.6,"identity":34.8,"society":47.1,"global":45.8},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"ensemble--complementary-mid--tdefault--tplcurrent--542975c7"}},"coverage":4,"index":39.1,"indexLo":34.5,"indexHi":43.7,"cost100":2.5837,"latencyP50":null,"flagged":0,"runs":4},{"key":"althing|qwen/qwen3.8-flash|default|current|||","base":"Qwen 3.8 Flash","name":"Qwen 3.8 Flash, persona","short":"Qwen 3.8 Flash","baseId":"qwen/qwen3.8-flash","members":[],"lab":"Alibaba","method":"althing","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.799456,"idx":11.4,"lo":-2,"hi":22.8,"pDist":0.729946,"pRank":0.690003,"pRefuse":0.978418,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":94.7828,"badge":"verified","holdoutGap":0.00852,"holdoutLimit":0.1066,"topicIdx":{"politics":10.7,"economy":58.6,"technology":34.2,"science-health":13.8,"identity":-45,"society":80.4,"global":12},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--qwen3.8-flash--tdefault--tplcurrent--abe04fc7"},"gss":{"sps":0.849052,"idx":43.2,"lo":35.3,"hi":51,"pDist":0.782206,"pRank":0.792898,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":60.2391,"badge":"verified","holdoutGap":0.010801,"holdoutLimit":0.129099,"topicIdx":{"politics":48.1,"economy":42.7,"technology":41.7,"science-health":68.7,"identity":22.2,"society":52.3,"global":31.1},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"althing--qwen3.8-flash--tdefault--tplcurrent--1b56d7bd"},"opinionsqa":{"sps":0.86609,"idx":43.6,"lo":36.1,"hi":50.4,"pDist":0.785939,"pRank":0.818058,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":70.9391,"badge":"verified","holdoutGap":0.041317,"holdoutLimit":0.104257,"topicIdx":{"politics":-19.8,"economy":8.6,"technology":44.2,"science-health":36.9,"identity":27.6,"society":62.4},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"althing--qwen3.8-flash--tdefault--tplcurrent--a4e37b94"},"subpop":{"sps":0.830041,"idx":30.5,"lo":20.6,"hi":38.8,"pDist":0.736691,"pRank":0.773814,"pRefuse":0.979618,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":82.925,"badge":"verified","holdoutGap":0.046758,"holdoutLimit":0.094491,"topicIdx":{"politics":22.5,"economy":62.9,"technology":24.5,"science-health":53.3,"identity":36,"society":30.3,"global":43.5},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--qwen3.8-flash--tdefault--tplcurrent--7e9168ad"}},"coverage":4,"index":32.2,"indexLo":27.5,"indexHi":36.9,"cost100":null,"latencyP50":77.22,"flagged":0,"runs":4},{"key":"raw|deepseek/deepseek-v4.1-flash|default|current|||","base":"DeepSeek V4.1 Flash","name":"DeepSeek V4.1 Flash, raw","short":"DeepSeek V4.1 Flash","baseId":"deepseek/deepseek-v4.1-flash","members":[],"lab":"DeepSeek","method":"raw","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.79398,"idx":8.9,"lo":-4.9,"hi":22.8,"pDist":0.706848,"pRank":0.695357,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":2.641,"badge":"verified","holdoutGap":0.0463,"holdoutLimit":0.1066,"topicIdx":{"politics":-8.3,"economy":37.3,"technology":34.1,"science-health":45.3,"identity":-8.2,"society":56.7,"global":12.2},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"raw--deepseek-v4.1-flash--tdefault--tplcurrent--3a0f2c7c"},"gss":{"sps":0.833673,"idx":37.3,"lo":29.3,"hi":45.1,"pDist":0.730937,"pRank":0.798033,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":2.484,"badge":"verified","holdoutGap":0.013581,"holdoutLimit":0.129099,"topicIdx":{"politics":41.8,"economy":42.1,"technology":29,"science-health":56.2,"identity":22.2,"society":41.8,"global":33.4},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"raw--deepseek-v4.1-flash--tdefault--tplcurrent--51ffcc3a"},"opinionsqa":{"sps":0.873957,"idx":46.9,"lo":41.3,"hi":52.7,"pDist":0.790707,"pRank":0.836893,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":2.359,"badge":"verified","holdoutGap":0.04174,"holdoutLimit":0.104257,"topicIdx":{"politics":39.3,"economy":20.3,"technology":42,"science-health":57.4,"identity":50.3,"society":53.8},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"raw--deepseek-v4.1-flash--tdefault--tplcurrent--9e8ad855"},"subpop":{"sps":0.83753,"idx":33.7,"lo":24.3,"hi":41.8,"pDist":0.739994,"pRank":0.792665,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":2.64,"badge":"verified","holdoutGap":0.022806,"holdoutLimit":0.094491,"topicIdx":{"politics":36.9,"economy":50,"technology":23.7,"science-health":59.2,"identity":25.4,"society":44.3,"global":39},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"raw--deepseek-v4.1-flash--tdefault--tplcurrent--a356727f"}},"coverage":4,"index":31.7,"indexLo":26.9,"indexHi":36.5,"cost100":null,"latencyP50":2.53,"flagged":0,"runs":4},{"key":"ensemble|3-model-blend|default|current|||","base":"Ensemble: Haiku 4.5, Gemini 2.5 Flash-Lite, GPT-4o mini (Althing)","name":"Ensemble of Haiku 4.5, Gemini 2.5 Flash-Lite, GPT-4o mini (all persona)","short":"Haiku + Gemini + GPT-4o mini, persona","baseId":"ensemble","members":[{"id":"anthropic/claude-haiku-4-5","method":"althing"},{"id":"google/gemini-2.5-flash-lite","method":"althing"},{"id":"openai/gpt-4o-mini","method":"althing"}],"lab":"DataViking","method":"ensemble","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.810184,"idx":16.2,"lo":3.9,"hi":27.8,"pDist":0.754855,"pRank":0.705566,"pRefuse":0.970131,"n":100,"runs":1,"samplesPerQuestion":null,"cost100":2.632413,"latencyP50":null,"badge":"verified","holdoutGap":0.053927,"holdoutLimit":0.1066,"topicIdx":{"politics":18.6,"economy":64.7,"technology":34.8,"science-health":49,"identity":-20.6,"society":87.2,"global":7.2},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"ensemble--3-model-blend--tdefault--tplcurrent--3ba7cfb3"},"gss":{"sps":0.816146,"idx":30.7,"lo":21.7,"hi":38.8,"pDist":0.729915,"pRank":0.746288,"pRefuse":0.972236,"n":75,"runs":1,"samplesPerQuestion":null,"cost100":2.368756,"latencyP50":null,"badge":"verified","holdoutGap":0.000328,"holdoutLimit":0.129099,"topicIdx":{"politics":44,"economy":12.4,"technology":47.2,"science-health":38.7,"identity":12.7,"society":34.9,"global":-11.4},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"ensemble--3-model-blend--tdefault--tplcurrent--3f093dfd"},"opinionsqa":{"sps":0.87124,"idx":45.8,"lo":38.7,"hi":53,"pDist":0.804849,"pRank":0.8146,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":null,"cost100":2.280196,"latencyP50":null,"badge":"verified","holdoutGap":0.004661,"holdoutLimit":0.104257,"topicIdx":{"politics":8.7,"economy":8.6,"technology":47.7,"science-health":40.9,"identity":40.8,"society":54.6},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"ensemble--3-model-blend--tdefault--tplcurrent--1bef3e62"},"subpop":{"sps":0.823337,"idx":27.7,"lo":20.8,"hi":34.4,"pDist":0.753574,"pRank":0.739771,"pRefuse":0.976666,"n":100,"runs":1,"samplesPerQuestion":null,"cost100":2.506444,"latencyP50":null,"badge":"verified","holdoutGap":0.046891,"holdoutLimit":0.094491,"topicIdx":{"politics":17.2,"economy":51.8,"technology":27.4,"science-health":46.7,"identity":29.2,"society":29.6,"global":35},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"ensemble--3-model-blend--tdefault--tplcurrent--ef75a647"}},"coverage":4,"index":30.1,"indexLo":25.7,"indexHi":34.5,"cost100":2.447,"latencyP50":null,"flagged":0,"runs":4},{"key":"raw|qwen/qwen3.8-flash|default|current|||","base":"Qwen 3.8 Flash","name":"Qwen 3.8 Flash, raw","short":"Qwen 3.8 Flash","baseId":"qwen/qwen3.8-flash","members":[],"lab":"Alibaba","method":"raw","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.793668,"idx":8.8,"lo":-4.3,"hi":20.5,"pDist":0.72741,"pRank":0.673861,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":8.218,"badge":"verified","holdoutGap":0.059952,"holdoutLimit":0.1066,"topicIdx":{"politics":8.6,"economy":30.1,"technology":34.4,"science-health":32,"identity":-11.2,"society":56.7,"global":2.7},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"raw--qwen3.8-flash--tdefault--tplcurrent--6c396a0d"},"gss":{"sps":0.83603,"idx":38.2,"lo":30.2,"hi":46,"pDist":0.757131,"pRank":0.77891,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":4.156,"badge":"verified","holdoutGap":0.001254,"holdoutLimit":0.129099,"topicIdx":{"politics":44.8,"economy":34,"technology":27.8,"science-health":52.8,"identity":30.2,"society":36.9,"global":20.8},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"raw--qwen3.8-flash--tdefault--tplcurrent--23b1d7af"},"opinionsqa":{"sps":0.858154,"idx":40.3,"lo":33.4,"hi":46.8,"pDist":0.78146,"pRank":0.798731,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":7.672,"badge":"verified","holdoutGap":0.004766,"holdoutLimit":0.104257,"topicIdx":{"politics":36.4,"economy":44.6,"technology":36.2,"science-health":50.3,"identity":25.5,"society":51.3},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"raw--qwen3.8-flash--tdefault--tplcurrent--e4b16fe3"},"subpop":{"sps":0.823478,"idx":27.8,"lo":20.6,"hi":34.8,"pDist":0.74347,"pRank":0.747034,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":4.265,"badge":"verified","holdoutGap":0.053732,"holdoutLimit":0.094491,"topicIdx":{"politics":32.7,"economy":57.6,"technology":13,"science-health":46.5,"identity":38.7,"society":36.7,"global":6.8},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"raw--qwen3.8-flash--tdefault--tplcurrent--b86a748e"}},"coverage":4,"index":28.8,"indexLo":24.4,"indexHi":33.2,"cost100":null,"latencyP50":6.08,"flagged":0,"runs":4},{"key":"althing|deepseek/deepseek-v4.1-flash|default|current|||","base":"DeepSeek V4.1 Flash","name":"DeepSeek V4.1 Flash, persona","short":"DeepSeek V4.1 Flash","baseId":"deepseek/deepseek-v4.1-flash","members":[],"lab":"DeepSeek","method":"althing","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.778229,"idx":1.9,"lo":-12.9,"hi":15.8,"pDist":0.660878,"pRank":0.694076,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":42.9843,"badge":"verified","holdoutGap":0.038836,"holdoutLimit":0.1066,"topicIdx":{"politics":-13.3,"economy":48.3,"technology":24.4,"science-health":48.5,"identity":-40,"society":56.7,"global":5.4},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--deepseek-v4.1-flash--tdefault--tplcurrent--578e68dd"},"gss":{"sps":0.822488,"idx":33.1,"lo":25.5,"hi":40.4,"pDist":0.709972,"pRank":0.785442,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":22.5594,"badge":"verified","holdoutGap":0.005877,"holdoutLimit":0.129099,"topicIdx":{"politics":43.9,"economy":36.4,"technology":23,"science-health":54.5,"identity":9.1,"society":40.3,"global":18.6},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"althing--deepseek-v4.1-flash--tdefault--tplcurrent--f5687570"},"opinionsqa":{"sps":0.831886,"idx":29.2,"lo":21.8,"hi":36.7,"pDist":0.716528,"pRank":0.784858,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":34.7656,"badge":"verified","holdoutGap":0.001838,"holdoutLimit":0.104257,"topicIdx":{"politics":-19.8,"economy":-11.8,"technology":30.9,"science-health":29.4,"identity":19.6,"society":39.8},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"althing--deepseek-v4.1-flash--tdefault--tplcurrent--6c5e905b"},"subpop":{"sps":0.815803,"idx":24.5,"lo":13.8,"hi":33.5,"pDist":0.690046,"pRank":0.777432,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":48.7016,"badge":"verified","holdoutGap":0.0149,"holdoutLimit":0.094491,"topicIdx":{"politics":21.3,"economy":29.2,"technology":21.9,"science-health":33.5,"identity":19,"society":26.1,"global":49.1},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--deepseek-v4.1-flash--tdefault--tplcurrent--9f683ba6"}},"coverage":4,"index":22.2,"indexLo":17.1,"indexHi":27.3,"cost100":null,"latencyP50":37.25,"flagged":0,"runs":4},{"key":"althing|anthropic/claude-haiku-4-5|0.3|current|||","base":"Claude Haiku 4.5","name":"Claude Haiku 4.5, persona (temperature 0.3)","short":"Haiku 4.5","baseId":"anthropic/claude-haiku-4-5","members":[],"lab":"Anthropic","method":"althing","variant":"temperature 0.3","harness":null,"ds":{"opinionsqa":{"sps":0.814418,"idx":21.8,"lo":13.9,"hi":29.4,"pDist":0.686546,"pRank":0.762436,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":2.125875,"latencyP50":85.3328,"badge":"verified","holdoutGap":0.0112,"holdoutLimit":0.104257,"topicIdx":{"politics":-19.8,"economy":-18.3,"technology":22.1,"science-health":21.8,"identity":18.9,"society":30.7},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"althing--claude-haiku-4.5--t0.3--tplcurrent--70eaffe7"}},"coverage":1,"index":21.8,"indexLo":14.1,"indexHi":29.6,"cost100":2.1259,"latencyP50":85.33,"flagged":0,"runs":1},{"key":"althing|google/gemini-3.5-flash-lite|default|current|||","base":"Gemini 3.5 Flash-Lite","name":"Gemini 3.5 Flash-Lite, persona","short":"Gemini 3.5 Flash-Lite","baseId":"google/gemini-3.5-flash-lite","members":[],"lab":"Google","method":"althing","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.77873,"idx":2.1,"lo":-10.6,"hi":16.2,"pDist":0.667233,"pRank":0.693325,"pRefuse":0.97563,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.05471,"latencyP50":21.9156,"badge":"verified","holdoutGap":0.079628,"holdoutLimit":0.1066,"topicIdx":{"politics":-2.7,"economy":37.6,"technology":5.5,"science-health":10.5,"identity":-54.8,"society":56.7,"global":10},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--gemini-3.5-flash-lite--tdefault--tplcurrent--09cfc7e5"},"gss":{"sps":0.814562,"idx":30.1,"lo":20.7,"hi":38.8,"pDist":0.695718,"pRank":0.775917,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":0.0442,"latencyP50":15.2829,"badge":"verified","holdoutGap":0.00591,"holdoutLimit":0.129099,"topicIdx":{"politics":43.1,"economy":5.7,"technology":26.7,"science-health":38.3,"identity":10.6,"society":42.2,"global":21.6},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"althing--gemini-3.5-flash-lite--tdefault--tplcurrent--dc95a0a6"},"opinionsqa":{"sps":0.83937,"idx":32.3,"lo":24.4,"hi":39.8,"pDist":0.734419,"pRank":0.789419,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.044558,"latencyP50":15.8547,"badge":"verified","holdoutGap":0.043846,"holdoutLimit":0.104257,"topicIdx":{"politics":60.7,"economy":-23,"technology":28.6,"science-health":24.6,"identity":30.8,"society":49.9},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"althing--gemini-3.5-flash-lite--tdefault--tplcurrent--20a55110"},"subpop":{"sps":0.810683,"idx":22.4,"lo":14,"hi":30.7,"pDist":0.6837,"pRank":0.768419,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.049004,"latencyP50":16.0922,"badge":"verified","holdoutGap":0.045774,"holdoutLimit":0.094491,"topicIdx":{"politics":27.6,"economy":40.7,"technology":20.5,"science-health":8.9,"identity":18.5,"society":17.1,"global":28.1},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--gemini-3.5-flash-lite--tdefault--tplcurrent--8892e804"}},"coverage":4,"index":21.7,"indexLo":16.8,"indexHi":26.6,"cost100":0.0481,"latencyP50":17.29,"flagged":0,"runs":4},{"key":"althing|anthropic/claude-haiku-4-5|default|structured|||","base":"Claude Haiku 4.5","name":"Claude Haiku 4.5, persona (structured template)","short":"Haiku 4.5","baseId":"anthropic/claude-haiku-4-5","members":[],"lab":"Anthropic","method":"althing","variant":"structured template","harness":null,"ds":{"gss":{"sps":0.791467,"idx":21.3,"lo":9.4,"hi":32.4,"pDist":0.666072,"pRank":0.736279,"pRefuse":0.972051,"n":75,"runs":2,"samplesPerQuestion":30,"cost100":null,"latencyP50":6.6938,"badge":"verified","holdoutGap":0.082696,"holdoutLimit":0.129099,"topicIdx":{"politics":44.2,"economy":23.4,"technology":23,"science-health":4.3,"identity":-3.8,"society":24.2,"global":25.7},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"synthpanel--claude-haiku-4-5--tdefault--tplstructured--fd829138"}},"coverage":1,"index":21.3,"indexLo":9.8,"indexHi":32.8,"cost100":null,"latencyP50":6.69,"flagged":0,"runs":2},{"key":"althing|anthropic/claude-sonnet-4|default|current|||","base":"Claude Sonnet 4","name":"Claude Sonnet 4, persona","short":"Sonnet 4","baseId":"anthropic/claude-sonnet-4","members":[],"lab":"Anthropic","method":"althing","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.768283,"idx":-2.6,"lo":-16.3,"hi":10.3,"pDist":0.651759,"pRank":0.673358,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":5.986725,"latencyP50":37.0532,"badge":"verified","holdoutGap":0.055118,"holdoutLimit":0.1066,"topicIdx":{"politics":-8.3,"economy":43.6,"technology":20.9,"science-health":30.6,"identity":-50.3,"society":56.7,"global":-3.4},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--claude-sonnet-4--tdefault--tplcurrent--e3d15dbd"},"gss":{"sps":0.812875,"idx":29.4,"lo":19.9,"hi":38.2,"pDist":0.695311,"pRank":0.771262,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":5.749,"latencyP50":22.5828,"badge":"verified","holdoutGap":0.017461,"holdoutLimit":0.129099,"topicIdx":{"politics":44.1,"economy":24,"technology":22.6,"science-health":60.9,"identity":-0.4,"society":38.8,"global":-14.9},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"althing--claude-sonnet-4--tdefault--tplcurrent--a74d2d89"},"opinionsqa":{"sps":0.841319,"idx":33.2,"lo":24.9,"hi":40.7,"pDist":0.731459,"pRank":0.798226,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":5.438685,"latencyP50":40.8625,"badge":"verified","holdoutGap":0.020572,"holdoutLimit":0.104257,"topicIdx":{"politics":-19.8,"economy":-32.8,"technology":35.7,"science-health":31.4,"identity":18.3,"society":50.3},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"althing--claude-sonnet-4--tdefault--tplcurrent--586e2e10"},"subpop":{"sps":0.803322,"idx":19.3,"lo":8.1,"hi":28.3,"pDist":0.666233,"pRank":0.763801,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":5.717445,"latencyP50":40.8625,"badge":"verified","holdoutGap":0.01873,"holdoutLimit":0.094491,"topicIdx":{"politics":7.4,"economy":29.2,"technology":22.5,"science-health":21.4,"identity":9.8,"society":27.3,"global":45.4},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--claude-sonnet-4--tdefault--tplcurrent--22a5d119"}},"coverage":4,"index":19.8,"indexLo":14.6,"indexHi":25,"cost100":5.723,"latencyP50":35.34,"flagged":0,"runs":4},{"key":"raw|google/gemini-3.5-flash-lite|default|current|||","base":"Gemini 3.5 Flash-Lite","name":"Gemini 3.5 Flash-Lite, raw","short":"Gemini 3.5 Flash-Lite","baseId":"google/gemini-3.5-flash-lite","members":[],"lab":"Google","method":"raw","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.769983,"idx":-1.8,"lo":-17.4,"hi":12.4,"pDist":0.663172,"pRank":0.685097,"pRefuse":0.961679,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.034636,"latencyP50":2.734,"badge":"verified","holdoutGap":0.055367,"holdoutLimit":0.1066,"topicIdx":{"politics":-24.6,"economy":47.5,"technology":15.5,"science-health":32.5,"identity":-45.6,"society":56.7,"global":9.4},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"raw--gemini-3.5-flash-lite--tdefault--tplcurrent--8aa5510f"},"gss":{"sps":0.807438,"idx":27.4,"lo":18.6,"hi":35.3,"pDist":0.70156,"pRank":0.748704,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":0.034365,"latencyP50":2.593,"badge":"verified","holdoutGap":0.045844,"holdoutLimit":0.129099,"topicIdx":{"politics":34.3,"economy":6.1,"technology":35.4,"science-health":37.4,"identity":18.4,"society":30.9,"global":0.4},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"raw--gemini-3.5-flash-lite--tdefault--tplcurrent--2e530450"},"opinionsqa":{"sps":0.836063,"idx":30.9,"lo":23.2,"hi":38.1,"pDist":0.729509,"pRank":0.78441,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.031634,"latencyP50":2.656,"badge":"verified","holdoutGap":0.048343,"holdoutLimit":0.104257,"topicIdx":{"politics":28,"economy":-2.7,"technology":27.6,"science-health":18.6,"identity":37.4,"society":42.9},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"raw--gemini-3.5-flash-lite--tdefault--tplcurrent--dad20c6a"},"subpop":{"sps":0.800542,"idx":18.1,"lo":9.7,"hi":26.7,"pDist":0.665597,"pRank":0.755822,"pRefuse":0.980207,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.032928,"latencyP50":2.641,"badge":"verified","holdoutGap":0.030371,"holdoutLimit":0.094491,"topicIdx":{"politics":24,"economy":59.8,"technology":13.5,"science-health":-2.5,"identity":-1.7,"society":38.6,"global":6.7},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"raw--gemini-3.5-flash-lite--tdefault--tplcurrent--30a43ab6"}},"coverage":4,"index":18.7,"indexLo":13.6,"indexHi":23.8,"cost100":0.0334,"latencyP50":2.66,"flagged":0,"runs":4},{"key":"althing|meta/muse-glimmer-30b|default|current|||","base":"Muse Glimmer 30B","name":"Muse Glimmer 30B, persona","short":"Muse Glimmer 30B","baseId":"meta/muse-glimmer-30b","members":[],"lab":"Meta","method":"althing","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.761592,"idx":-5.6,"lo":-21,"hi":8.4,"pDist":0.636427,"pRank":0.67212,"pRefuse":0.97623,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":77.1796,"badge":"verified","holdoutGap":0.039745,"holdoutLimit":0.1066,"topicIdx":{"politics":-21.2,"economy":53,"technology":17.6,"science-health":6.2,"identity":-46.1,"society":56.7,"global":0.1},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--muse-glimmer-30b--tdefault--tplcurrent--cb979206"},"gss":{"sps":0.804997,"idx":26.4,"lo":18.2,"hi":34.4,"pDist":0.696179,"pRank":0.747155,"pRefuse":0.971658,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":86.2937,"badge":"verified","holdoutGap":0.03283,"holdoutLimit":0.129099,"topicIdx":{"politics":42,"economy":26.4,"technology":20.3,"science-health":31.8,"identity":4.6,"society":29.6,"global":21.6},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"althing--muse-glimmer-30b--tdefault--tplcurrent--0c88b48a"},"opinionsqa":{"sps":0.832725,"idx":29.5,"lo":22,"hi":36.7,"pDist":0.716334,"pRank":0.78757,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":62.2062,"badge":"verified","holdoutGap":0.015323,"holdoutLimit":0.104257,"topicIdx":{"politics":-19.8,"economy":-22.3,"technology":31.6,"science-health":23.2,"identity":16.7,"society":45.6},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"althing--muse-glimmer-30b--tdefault--tplcurrent--a4ad6ce2"},"subpop":{"sps":0.809188,"idx":21.8,"lo":10.9,"hi":30.4,"pDist":0.678194,"pRank":0.769439,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":56.8422,"badge":"verified","holdoutGap":0.050689,"holdoutLimit":0.094491,"topicIdx":{"politics":4.7,"economy":29.2,"technology":22.6,"science-health":37.4,"identity":26.7,"society":25.6,"global":47.3},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--muse-glimmer-30b--tdefault--tplcurrent--69d0c98e"}},"coverage":4,"index":18,"indexLo":12.8,"indexHi":23.2,"cost100":null,"latencyP50":70.63,"flagged":0,"runs":4},{"key":"althing|anthropic/claude-haiku-5-5|default|structured||global-respondents|","base":"Claude Haiku 5.5","name":"Claude Haiku 5.5, persona (structured template, global-respondents personas)","short":"Haiku 5.5","baseId":"anthropic/claude-haiku-5-5","members":[],"lab":"Anthropic","method":"althing","variant":"structured template, global-respondents personas","harness":null,"ds":{"globalopinionqa":{"sps":0.810688,"idx":16.4,"lo":4.4,"hi":27.6,"pDist":0.725962,"pRank":0.726368,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.634845,"latencyP50":61,"badge":"verified","holdoutGap":0.045367,"holdoutLimit":0.1066,"topicIdx":{"politics":6.2,"economy":55.8,"technology":28.4,"science-health":14.5,"identity":14.3,"society":56.7,"global":16.4},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--claude-haiku-5.5--tdefault--tplstructured--6baec9b6"}},"coverage":1,"index":16.4,"indexLo":4.8,"indexHi":28,"cost100":0.6348,"latencyP50":61,"flagged":0,"runs":1},{"key":"althing|openai/gpt-6-luna|default|current|||","base":"GPT-6 Luna","name":"GPT-6 Luna, persona","short":"GPT-6 Luna","baseId":"openai/gpt-6-luna","members":[],"lab":"OpenAI","method":"althing","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.760212,"idx":-6.2,"lo":-21.7,"hi":7.8,"pDist":0.634735,"pRank":0.666168,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":94.1563,"badge":"verified","holdoutGap":0.071141,"holdoutLimit":0.1066,"topicIdx":{"politics":-26,"economy":43.4,"technology":17.6,"science-health":9.2,"identity":-51.4,"society":56.7,"global":4.4},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--gpt-6-luna--tdefault--tplcurrent--1425531c"},"gss":{"sps":0.801406,"idx":25.1,"lo":15.5,"hi":33.7,"pDist":0.691119,"pRank":0.741048,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":56.0438,"badge":"verified","holdoutGap":0.020364,"holdoutLimit":0.129099,"topicIdx":{"politics":39.2,"economy":8.6,"technology":21.7,"science-health":36.7,"identity":3.1,"society":28.1,"global":48.6},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"althing--gpt-6-luna--tdefault--tplcurrent--cabf3183"},"opinionsqa":{"sps":0.829359,"idx":28.1,"lo":19.7,"hi":36.2,"pDist":0.72081,"pRank":0.772997,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":82.6328,"badge":"verified","holdoutGap":0.005464,"holdoutLimit":0.104257,"topicIdx":{"politics":-19.8,"economy":-8.2,"technology":32.1,"science-health":26.1,"identity":18.7,"society":35},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"althing--gpt-6-luna--tdefault--tplcurrent--1c0908e4"},"subpop":{"sps":0.798292,"idx":17.2,"lo":5.9,"hi":25.4,"pDist":0.665369,"pRank":0.749577,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":83.4938,"badge":"verified","holdoutGap":0.01805,"holdoutLimit":0.094491,"topicIdx":{"politics":3.4,"economy":29.2,"technology":19.2,"science-health":31.6,"identity":8,"society":26.7,"global":42.8},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--gpt-6-luna--tdefault--tplcurrent--d010082e"}},"coverage":4,"index":16.1,"indexLo":10.7,"indexHi":21.5,"cost100":null,"latencyP50":79.08,"flagged":0,"runs":4},{"key":"althing|openai/gpt-6-luna|default|structured||global-respondents|codex","base":"GPT-6 Luna","name":"GPT-6 Luna via codex, persona (structured template, global-respondents personas)","short":"GPT-6 Luna via codex","baseId":"openai/gpt-6-luna","members":[],"lab":"OpenAI","method":"althing","variant":"structured template, global-respondents personas","harness":"codex","ds":{"globalopinionqa":{"sps":0.808885,"idx":15.6,"lo":3.6,"hi":27.4,"pDist":0.747379,"pRank":0.699545,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":180.125,"badge":"verified","holdoutGap":0.000621,"holdoutLimit":0.1066,"topicIdx":{"politics":3,"economy":68.7,"technology":33.2,"science-health":35.5,"identity":-20.7,"society":56.7,"global":19.3},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--gpt-6-luna--tdefault--tplstructured--74032abc"}},"coverage":1,"index":15.6,"indexLo":3.7,"indexHi":27.5,"cost100":null,"latencyP50":180.13,"flagged":0,"runs":1},{"key":"althing|anthropic/claude-haiku-5-5|default|current|||","base":"Claude Haiku 5.5","name":"Claude Haiku 5.5, persona","short":"Haiku 5.5","baseId":"anthropic/claude-haiku-5-5","members":[],"lab":"Anthropic","method":"althing","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.774391,"idx":0.2,"lo":-14.3,"hi":12.7,"pDist":0.660135,"pRank":0.684175,"pRefuse":0.978863,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":3.23577,"latencyP50":32.4687,"badge":"verified","holdoutGap":0.092102,"holdoutLimit":0.1066,"topicIdx":{"politics":5.1,"economy":32.2,"technology":15.6,"science-health":-8.1,"identity":-45.3,"society":56.7,"global":-0.9},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--claude-haiku-5.5--tdefault--tplcurrent--c3b8c46a"},"gss":{"sps":0.80295,"idx":25.7,"lo":15,"hi":35,"pDist":0.691681,"pRank":0.749066,"pRefuse":0.968102,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":2.49346,"latencyP50":16.3828,"badge":"verified","holdoutGap":0.008516,"holdoutLimit":0.129099,"topicIdx":{"politics":36.4,"economy":-5,"technology":34.5,"science-health":51.4,"identity":3,"society":30.7,"global":33.9},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"althing--claude-haiku-5.5--tdefault--tplcurrent--39c1180c"},"opinionsqa":{"sps":0.822803,"idx":25.3,"lo":17.3,"hi":32.7,"pDist":0.700423,"pRank":0.775571,"pRefuse":0.992416,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":1.82173,"latencyP50":29.5141,"badge":"verified","holdoutGap":0.039736,"holdoutLimit":0.104257,"topicIdx":{"politics":52,"economy":-3,"technology":24.9,"science-health":21.1,"identity":25,"society":31.2},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"althing--claude-haiku-5.5--tdefault--tplcurrent--b3b22d00"},"subpop":{"sps":0.783362,"idx":10.9,"lo":2.1,"hi":19.2,"pDist":0.645295,"pRank":0.730853,"pRefuse":0.973938,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":2.47506,"latencyP50":28.2485,"badge":"verified","holdoutGap":0.009569,"holdoutLimit":0.094491,"topicIdx":{"politics":-13,"economy":43.4,"technology":20.5,"science-health":6.9,"identity":6.7,"society":19.7,"global":28.6},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--claude-haiku-5.5--tdefault--tplcurrent--c9b6a625"}},"coverage":4,"index":15.5,"indexLo":10.4,"indexHi":20.6,"cost100":2.5065,"latencyP50":26.65,"flagged":0,"runs":4},{"key":"althing|openai/gpt-6-luna|default|current|||codex","base":"GPT-6 Luna","name":"GPT-6 Luna via codex, persona","short":"GPT-6 Luna via codex","baseId":"openai/gpt-6-luna","members":[],"lab":"OpenAI","method":"althing","variant":null,"harness":"codex","ds":{"globalopinionqa":{"sps":0.762266,"idx":-5.3,"lo":-20,"hi":8.5,"pDist":0.665298,"pRank":0.641767,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":122.4063,"badge":"verified","holdoutGap":0.013355,"holdoutLimit":0.1066,"topicIdx":{"politics":-21.7,"economy":43,"technology":21.8,"science-health":-16.3,"identity":-34.4,"society":88.4,"global":1.4},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--gpt-6-luna--tdefault--tplcurrent--59b62c0c"},"gss":{"sps":0.782199,"idx":17.8,"lo":5.7,"hi":29.7,"pDist":0.691045,"pRank":0.683502,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":111.8562,"badge":"verified","holdoutGap":0.014394,"holdoutLimit":0.129099,"topicIdx":{"politics":34.3,"economy":-28,"technology":37.2,"science-health":23.4,"identity":1.1,"society":16.4,"global":29.2},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"althing--gpt-6-luna--tdefault--tplcurrent--4603eaf6"},"opinionsqa":{"sps":0.832773,"idx":29.5,"lo":22.2,"hi":36.6,"pDist":0.745826,"pRank":0.758221,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":122.089,"badge":"verified","holdoutGap":0.059624,"holdoutLimit":0.104257,"topicIdx":{"politics":-19.8,"economy":22.6,"technology":34.7,"science-health":21.2,"identity":16.6,"society":34.4},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"althing--gpt-6-luna--tdefault--tplcurrent--3ccc3a6d"},"subpop":{"sps":0.802001,"idx":18.7,"lo":8.5,"hi":27.4,"pDist":0.696016,"pRank":0.730057,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":123.9875,"badge":"verified","holdoutGap":0.041054,"holdoutLimit":0.094491,"topicIdx":{"politics":5.9,"economy":27.7,"technology":26.5,"science-health":16.6,"identity":18,"society":8.7,"global":47.7},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--gpt-6-luna--tdefault--tplcurrent--956505b1"}},"coverage":4,"index":15.2,"indexLo":9.7,"indexHi":20.7,"cost100":null,"latencyP50":120.08,"flagged":0,"runs":4},{"key":"raw|google/gemini-2.5-flash-lite|default|current|||","base":"Gemini 2.5 Flash-Lite","name":"Gemini 2.5 Flash-Lite, raw","short":"Gemini 2.5 Flash-Lite","baseId":"google/gemini-2.5-flash-lite","members":[],"lab":"Google","method":"raw","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.761344,"idx":-5.7,"lo":-19.4,"hi":8,"pDist":0.67856,"pRank":0.626508,"pRefuse":0.978965,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.034568,"latencyP50":1.25,"badge":"verified","holdoutGap":0.074717,"holdoutLimit":0.1066,"topicIdx":{"politics":-19.8,"economy":14.1,"technology":8.5,"science-health":28.8,"identity":-4.3,"society":56.7,"global":-6},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"raw--gemini-2.5-flash-lite--tdefault--tplcurrent--2c7b8ae9"},"gss":{"sps":0.813499,"idx":29.7,"lo":18.8,"hi":40.7,"pDist":0.719672,"pRank":0.749219,"pRefuse":0.971606,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":0.034344,"latencyP50":1.25,"badge":"verified","holdoutGap":0.031929,"holdoutLimit":0.129099,"topicIdx":{"politics":44.5,"economy":10.8,"technology":59.8,"science-health":52.6,"identity":13.8,"society":6.4,"global":22},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"raw--gemini-2.5-flash-lite--tdefault--tplcurrent--ae412982"},"opinionsqa":{"sps":0.825897,"idx":26.6,"lo":17.7,"hi":35.7,"pDist":0.73653,"pRank":0.746888,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.031596,"latencyP50":1.203,"badge":"verified","holdoutGap":0.007788,"holdoutLimit":0.104257,"topicIdx":{"politics":24.5,"economy":-4.7,"technology":21.8,"science-health":19.3,"identity":7.3,"society":53.7},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"raw--gemini-2.5-flash-lite--tdefault--tplcurrent--83919e22"},"subpop":{"sps":0.77833,"idx":8.8,"lo":-3.7,"hi":19.8,"pDist":0.653427,"pRank":0.701631,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.032886,"latencyP50":1.344,"badge":"verified","holdoutGap":0.086916,"holdoutLimit":0.094491,"topicIdx":{"politics":-11.1,"economy":12.1,"technology":18.3,"science-health":19.1,"identity":14,"society":4,"global":28.6},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"raw--gemini-2.5-flash-lite--tdefault--tplcurrent--5d34baac"}},"coverage":4,"index":14.9,"indexLo":9.2,"indexHi":20.6,"cost100":0.0333,"latencyP50":1.26,"flagged":0,"runs":4},{"key":"althing|google/gemini-2.5-flash-lite|0.3|current|||","base":"Gemini 2.5 Flash-Lite","name":"Gemini 2.5 Flash-Lite, persona (temperature 0.3)","short":"Gemini 2.5 Flash-Lite","baseId":"google/gemini-2.5-flash-lite","members":[],"lab":"Google","method":"althing","variant":"temperature 0.3","harness":null,"ds":{"opinionsqa":{"sps":0.7957,"idx":13.9,"lo":5.8,"hi":22.7,"pDist":0.654663,"pRank":0.738165,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.042194,"latencyP50":52.7563,"badge":"verified","holdoutGap":0.021303,"holdoutLimit":0.104257,"topicIdx":{"politics":-19.8,"economy":-23,"technology":12.5,"science-health":9.4,"identity":13.6,"society":25.1},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"althing--gemini-2.5-flash-lite--t0.3--tplcurrent--d771901f"}},"coverage":1,"index":13.9,"indexLo":5.5,"indexHi":22.4,"cost100":0.0422,"latencyP50":52.76,"flagged":0,"runs":1},{"key":"althing|anthropic/claude-sonnet-4-6|default|current|||","base":"Claude Sonnet 4.6","name":"Claude Sonnet 4.6, persona","short":"Sonnet 4.6","baseId":"anthropic/claude-sonnet-4-6","members":[],"lab":"Anthropic","method":"althing","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.758637,"idx":-6.9,"lo":-21.1,"hi":6.3,"pDist":0.622801,"pRank":0.673377,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":6.084795,"latencyP50":47.5609,"badge":"verified","holdoutGap":0.047416,"holdoutLimit":0.1066,"topicIdx":{"politics":-16.4,"economy":34.9,"technology":17.6,"science-health":-11,"identity":-58.5,"society":56.7,"global":0.7},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--claude-sonnet-4.6--tdefault--tplcurrent--3bd552af"},"gss":{"sps":0.813865,"idx":29.8,"lo":20,"hi":39,"pDist":0.688738,"pRank":0.780806,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":5.70836,"latencyP50":29.0546,"badge":"verified","holdoutGap":0.004937,"holdoutLimit":0.129099,"topicIdx":{"politics":44,"economy":5.7,"technology":36.9,"science-health":54,"identity":7.9,"society":31.8,"global":0.4},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"althing--claude-sonnet-4.6--tdefault--tplcurrent--5494841a"},"opinionsqa":{"sps":0.807349,"idx":18.8,"lo":10.9,"hi":26.1,"pDist":0.675492,"pRank":0.752282,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":5.23587,"latencyP50":71.4453,"badge":"verified","holdoutGap":0.052227,"holdoutLimit":0.104257,"topicIdx":{"politics":-19.8,"economy":-1.1,"technology":9.9,"science-health":21.9,"identity":24.7,"society":33.9},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"althing--claude-sonnet-4.6--tdefault--tplcurrent--1cec9e61"},"subpop":{"sps":0.782049,"idx":10.4,"lo":0,"hi":19.4,"pDist":0.632653,"pRank":0.733564,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":6.011115,"latencyP50":41.7829,"badge":"verified","holdoutGap":0.04088,"holdoutLimit":0.094491,"topicIdx":{"politics":7.4,"economy":34.9,"technology":7.9,"science-health":17.9,"identity":10.4,"society":12.3,"global":9.6},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--claude-sonnet-4.6--tdefault--tplcurrent--c6249a6f"}},"coverage":4,"index":13,"indexLo":7.8,"indexHi":18.2,"cost100":5.76,"latencyP50":47.46,"flagged":0,"runs":4},{"key":"raw|openai/gpt-4o|default|current|||","base":"GPT-4o","name":"GPT-4o, raw","short":"GPT-4o","baseId":"openai/gpt-4o","members":[],"lab":"OpenAI","method":"raw","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.764664,"idx":-4.2,"lo":-19.4,"hi":9.6,"pDist":0.686438,"pRank":0.629239,"pRefuse":0.978315,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.8926,"latencyP50":3.562,"badge":"verified","holdoutGap":0.026911,"holdoutLimit":0.1066,"topicIdx":{"politics":-17.7,"economy":36,"technology":34.9,"science-health":16.3,"identity":-27.1,"society":56.7,"global":-4.5},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"raw--gpt-4o--tdefault--tplcurrent--85667f87"},"gss":{"sps":0.802365,"idx":25.4,"lo":14.1,"hi":35.8,"pDist":0.704855,"pRank":0.730188,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":0.85908,"latencyP50":3.563,"badge":"verified","holdoutGap":0.019048,"holdoutLimit":0.129099,"topicIdx":{"politics":48.2,"economy":9.7,"technology":56.4,"science-health":30.1,"identity":2.8,"society":10.5,"global":0.4},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"raw--gpt-4o--tdefault--tplcurrent--3ea2ce7d"},"opinionsqa":{"sps":0.825484,"idx":26.5,"lo":17.1,"hi":36.3,"pDist":0.748634,"pRank":0.733547,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.8172,"latencyP50":3.343,"badge":"verified","holdoutGap":0.013356,"holdoutLimit":0.104257,"topicIdx":{"politics":40.1,"economy":-33.7,"technology":33.5,"science-health":37,"identity":14.9,"society":25.9},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"raw--gpt-4o--tdefault--tplcurrent--93c0c12c"},"subpop":{"sps":0.76327,"idx":2.5,"lo":-8.9,"hi":12.5,"pDist":0.662948,"pRank":0.646931,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.839445,"latencyP50":3.282,"badge":"verified","holdoutGap":0.00598,"holdoutLimit":0.094491,"topicIdx":{"politics":-1.5,"economy":-41.1,"technology":12.8,"science-health":-24.4,"identity":11.5,"society":-11.3,"global":19.6},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"raw--gpt-4o--tdefault--tplcurrent--3f1d7440"}},"coverage":4,"index":12.5,"indexLo":6.7,"indexHi":18.3,"cost100":0.8521,"latencyP50":3.44,"flagged":0,"runs":4},{"key":"althing|meta-llama/llama-3.3-70b-instruct|default|current|||","base":"Llama 3.3 70B","name":"Llama 3.3 70B, persona","short":"Llama 3.3 70B","baseId":"meta-llama/llama-3.3-70b-instruct","members":[],"lab":"Meta","method":"althing","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.763864,"idx":-4.5,"lo":-17.8,"hi":8.4,"pDist":0.630948,"pRank":0.680911,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.09853,"latencyP50":35.45,"badge":"verified","holdoutGap":0.01344,"holdoutLimit":0.1066,"topicIdx":{"politics":-6.1,"economy":10.4,"technology":37.8,"science-health":40.2,"identity":-32.9,"society":56.7,"global":-12.3},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--llama-3.3-70b-instruct--tdefault--tplcurrent--d7a23220"},"gss":{"sps":0.81393,"idx":29.8,"lo":21.3,"hi":37.7,"pDist":0.687362,"pRank":0.782378,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":0.090245,"latencyP50":15.6375,"badge":"verified","holdoutGap":0.019476,"holdoutLimit":0.129099,"topicIdx":{"politics":45.3,"economy":12,"technology":9.6,"science-health":55.4,"identity":7.2,"society":37.4,"global":5.1},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"althing--llama-3.3-70b-instruct--tdefault--tplcurrent--e0c2d6f9"},"opinionsqa":{"sps":0.802667,"idx":16.8,"lo":8.8,"hi":24.5,"pDist":0.675679,"pRank":0.738052,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.091608,"latencyP50":25.9407,"badge":"verified","holdoutGap":0.019565,"holdoutLimit":0.104257,"topicIdx":{"politics":-19.8,"economy":-23.9,"technology":18.2,"science-health":18,"identity":21.1,"society":19.2},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"althing--llama-3.3-70b-instruct--tdefault--tplcurrent--b48b9ede"},"subpop":{"sps":0.775725,"idx":7.7,"lo":-1.1,"hi":15.6,"pDist":0.623455,"pRank":0.723788,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.089885,"latencyP50":19.7157,"badge":"verified","holdoutGap":0.027328,"holdoutLimit":0.094491,"topicIdx":{"politics":22.9,"economy":-24.7,"technology":9,"science-health":24,"identity":-8.9,"society":2.5,"global":2.8},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--llama-3.3-70b-instruct--tdefault--tplcurrent--7df63c05"}},"coverage":4,"index":12.5,"indexLo":7.7,"indexHi":17.3,"cost100":0.0926,"latencyP50":24.19,"flagged":0,"runs":4},{"key":"althing|openai/gpt-4o-mini|0.3|current|||","base":"GPT-4o mini","name":"GPT-4o mini, persona (temperature 0.3)","short":"GPT-4o mini","baseId":"openai/gpt-4o-mini","members":[],"lab":"OpenAI","method":"althing","variant":"temperature 0.3","harness":null,"ds":{"opinionsqa":{"sps":0.791967,"idx":12.3,"lo":3.2,"hi":20.9,"pDist":0.649077,"pRank":0.732552,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.109276,"latencyP50":77.7172,"badge":"verified","holdoutGap":0.012033,"holdoutLimit":0.104257,"topicIdx":{"politics":11.7,"economy":-23,"technology":13,"science-health":0.9,"identity":20.2,"society":16.4},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"althing--gpt-4o-mini--t0.3--tplcurrent--a38abbba"}},"coverage":1,"index":12.3,"indexLo":3.5,"indexHi":21.2,"cost100":0.1093,"latencyP50":77.72,"flagged":0,"runs":1},{"key":"althing|anthropic/claude-haiku-5-5|default|structured|||","base":"Claude Haiku 5.5","name":"Claude Haiku 5.5, persona (structured template)","short":"Haiku 5.5","baseId":"anthropic/claude-haiku-5-5","members":[],"lab":"Anthropic","method":"althing","variant":"structured template","harness":null,"ds":{"globalopinionqa":{"sps":0.75311,"idx":-9.4,"lo":-23.4,"hi":4,"pDist":0.609389,"pRank":0.670208,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.636985,"latencyP50":46.7453,"badge":"verified","holdoutGap":0.073759,"holdoutLimit":0.1066,"topicIdx":{"politics":-13.4,"economy":27.6,"technology":17.6,"science-health":18.9,"identity":-25.7,"society":56.7,"global":-16.9},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--claude-haiku-5.5--tdefault--tplstructured--01b3507a"},"opinionsqa":{"sps":0.834444,"idx":30.3,"lo":23.5,"hi":36.9,"pDist":0.718182,"pRank":0.790879,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.62946,"latencyP50":46.5141,"badge":"verified","holdoutGap":0.026379,"holdoutLimit":0.104257,"topicIdx":{"politics":-19.8,"economy":-17.5,"technology":31.2,"science-health":15.2,"identity":38.6,"society":38.9},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"althing--claude-haiku-5.5--tdefault--tplstructured--f2973320"},"subpop":{"sps":0.790644,"idx":14,"lo":3.2,"hi":22.6,"pDist":0.635892,"pRank":0.756109,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.650175,"latencyP50":46.6437,"badge":"verified","holdoutGap":0.04642,"holdoutLimit":0.094491,"topicIdx":{"politics":3.7,"economy":29.2,"technology":19.2,"science-health":5.4,"identity":-3.9,"society":33.1,"global":23.7},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--claude-haiku-5.5--tdefault--tplstructured--0ea245df"}},"coverage":3,"index":11.6,"indexLo":5.6,"indexHi":17.6,"cost100":0.6389,"latencyP50":46.63,"flagged":0,"runs":3},{"key":"althing|openai/gpt-6-luna|default|structured|||codex","base":"GPT-6 Luna","name":"GPT-6 Luna via codex, persona (structured template)","short":"GPT-6 Luna via codex","baseId":"openai/gpt-6-luna","members":[],"lab":"OpenAI","method":"althing","variant":"structured template","harness":"codex","ds":{"globalopinionqa":{"sps":0.76475,"idx":-4.2,"lo":-18.8,"hi":10.4,"pDist":0.653872,"pRank":0.660646,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":136.8734,"badge":"verified","holdoutGap":0.036286,"holdoutLimit":0.1066,"topicIdx":{"politics":-20.7,"economy":42,"technology":27.2,"science-health":5,"identity":-49.8,"society":88.9,"global":3.4},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--gpt-6-luna--tdefault--tplstructured--c4cb9735"},"opinionsqa":{"sps":0.815631,"idx":22.3,"lo":12.3,"hi":31.2,"pDist":0.728301,"pRank":0.72432,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":124.1453,"badge":"verified","holdoutGap":0.063676,"holdoutLimit":0.104257,"topicIdx":{"politics":-37.8,"economy":10.1,"technology":34.6,"science-health":20,"identity":7.9,"society":15},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"althing--gpt-6-luna--tdefault--tplstructured--5fe413ba"},"subpop":{"sps":0.790997,"idx":14.1,"lo":3.1,"hi":24.3,"pDist":0.666088,"pRank":0.726974,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":127.5219,"badge":"verified","holdoutGap":0.052944,"holdoutLimit":0.094491,"topicIdx":{"politics":8.5,"economy":-13.6,"technology":20.6,"science-health":6.3,"identity":15.5,"society":2.3,"global":46.7},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--gpt-6-luna--tdefault--tplstructured--f0bd267a"}},"coverage":3,"index":10.7,"indexLo":3.9,"indexHi":17.5,"cost100":null,"latencyP50":129.51,"flagged":0,"runs":3},{"key":"althing|openai/gpt-4o-mini|default|current|||","base":"GPT-4o mini","name":"GPT-4o mini, persona","short":"GPT-4o mini","baseId":"openai/gpt-4o-mini","members":[],"lab":"OpenAI","method":"althing","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.778334,"idx":1.9,"lo":-12,"hi":14.9,"pDist":0.666942,"pRank":0.692881,"pRefuse":0.97518,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.127988,"latencyP50":32.8281,"badge":"verified","holdoutGap":0.012961,"holdoutLimit":0.1066,"topicIdx":{"politics":0.4,"economy":59.6,"technology":24.4,"science-health":37.7,"identity":-30.1,"society":56.7,"global":-7.1},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--gpt-4o-mini--tdefault--tplcurrent--f99ce6e2"},"gss":{"sps":0.801064,"idx":24.9,"lo":14.3,"hi":34.6,"pDist":0.675254,"pRank":0.755886,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":0.114572,"latencyP50":18.5297,"badge":"verified","holdoutGap":0.021323,"holdoutLimit":0.129099,"topicIdx":{"politics":41,"economy":16.8,"technology":37.7,"science-health":51.4,"identity":-5,"society":21.2,"global":21.6},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"althing--gpt-4o-mini--tdefault--tplcurrent--298bc463"},"subpop":{"sps":0.760117,"idx":1.1,"lo":-7.9,"hi":9.5,"pDist":0.612491,"pRank":0.687931,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.125005,"latencyP50":28.6046,"badge":"verified","holdoutGap":0.039244,"holdoutLimit":0.094491,"topicIdx":{"politics":12.1,"economy":11.1,"technology":6.9,"science-health":-20.9,"identity":-17.3,"society":-2.6,"global":-9.5},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--gpt-4o-mini--tdefault--tplcurrent--56114dbf"}},"coverage":3,"index":9.3,"indexLo":3,"indexHi":15.6,"cost100":0.1225,"latencyP50":26.65,"flagged":0,"runs":3},{"key":"althing|anthropic/claude-haiku-4-5|default|current|||","base":"Claude Haiku 4.5","name":"Claude Haiku 4.5, persona","short":"Haiku 4.5","baseId":"anthropic/claude-haiku-4-5","members":[],"lab":"Anthropic","method":"althing","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.761736,"idx":-5.5,"lo":-19.2,"hi":8.8,"pDist":0.650597,"pRank":0.68228,"pRefuse":0.95233,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":2.458955,"latencyP50":32.2906,"badge":"verified","holdoutGap":0.060773,"holdoutLimit":0.1066,"topicIdx":{"politics":-10.2,"economy":46.7,"technology":11.7,"science-health":23,"identity":-50.7,"society":-9.2,"global":-5.6},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--claude-haiku-4.5--tdefault--tplcurrent--071a33f3"},"gss":{"sps":0.782395,"idx":17.8,"lo":6.8,"hi":27.6,"pDist":0.65828,"pRank":0.716298,"pRefuse":0.972606,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":2.210127,"latencyP50":19.1218,"badge":"verified","holdoutGap":0.050721,"holdoutLimit":0.129099,"topicIdx":{"politics":34.3,"economy":5.7,"technology":15.6,"science-health":4.3,"identity":-0.9,"society":32.5,"global":0.4},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"althing--claude-haiku-4.5--tdefault--tplcurrent--c22c8e18"},"subpop":{"sps":0.770122,"idx":5.3,"lo":-5.3,"hi":14.1,"pDist":0.633509,"pRank":0.710912,"pRefuse":0.965945,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":2.358395,"latencyP50":31.0828,"badge":"verified","holdoutGap":0.02948,"holdoutLimit":0.094491,"topicIdx":{"politics":-13,"economy":40.1,"technology":19.2,"science-health":26.3,"identity":-15.4,"society":3.6,"global":25.6},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--claude-haiku-4.5--tdefault--tplcurrent--28536223"}},"coverage":3,"index":5.9,"indexLo":-0.8,"indexHi":12.6,"cost100":2.3425,"latencyP50":27.5,"flagged":0,"runs":3},{"key":"raw|meta-llama/llama-3.3-70b-instruct|default|current|||","base":"Llama 3.3 70B","name":"Llama 3.3 70B, raw","short":"Llama 3.3 70B","baseId":"meta-llama/llama-3.3-70b-instruct","members":[],"lab":"Meta","method":"raw","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.770055,"idx":-1.8,"lo":-15.9,"hi":11.8,"pDist":0.638848,"pRank":0.691584,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.052923,"latencyP50":2.36,"badge":"verified","holdoutGap":0.015238,"holdoutLimit":0.1066,"topicIdx":{"politics":-5.7,"economy":17.2,"technology":20.1,"science-health":29.9,"identity":-33.8,"society":56.7,"global":-3.1},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"raw--llama-3.3-70b-instruct--tdefault--tplcurrent--e4722076"},"gss":{"sps":0.814656,"idx":30.1,"lo":20.2,"hi":39.2,"pDist":0.696118,"pRank":0.7758,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":0.052109,"latencyP50":2.375,"badge":"verified","holdoutGap":0.036227,"holdoutLimit":0.129099,"topicIdx":{"politics":50.4,"economy":6,"technology":15.6,"science-health":46,"identity":11.1,"society":30,"global":0.4},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"raw--llama-3.3-70b-instruct--tdefault--tplcurrent--bb2f92c9"},"opinionsqa":{"sps":0.822591,"idx":25.2,"lo":17.4,"hi":33.4,"pDist":0.702964,"pRank":0.770538,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.046085,"latencyP50":2.937,"badge":"verified","holdoutGap":0.034463,"holdoutLimit":0.104257,"topicIdx":{"politics":-19.8,"economy":-36.4,"technology":19.2,"science-health":32.1,"identity":29.4,"society":41.7},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"raw--llama-3.3-70b-instruct--tdefault--tplcurrent--14030c06"},"subpop":{"sps":0.670135,"idx":-36.7,"lo":-43.5,"hi":-30.9,"pDist":0.650402,"pRank":0.742353,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":15,"cost100":0.024169,"latencyP50":1.156,"badge":"verified","holdoutGap":0.001405,"holdoutLimit":0.094491,"topicIdx":{"politics":-40.2,"economy":-25.1,"technology":-35.6,"science-health":-27.7,"identity":-41.9,"society":-26.4,"global":-49.9},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"raw--llama-3.3-70b-instruct--tdefault--tplcurrent--915d6808"}},"coverage":4,"index":4.2,"indexLo":-0.7,"indexHi":9.1,"cost100":0.0438,"latencyP50":2.21,"flagged":0,"runs":4},{"key":"althing|anthropic/claude-sonnet-5-5|default|current|||","base":"Claude Sonnet 5.5","name":"Claude Sonnet 5.5, persona","short":"Sonnet 5.5","baseId":"anthropic/claude-sonnet-5-5","members":[],"lab":"Anthropic","method":"althing","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.765404,"idx":-3.9,"lo":-17.8,"hi":9,"pDist":0.643397,"pRank":0.679678,"pRefuse":0.973138,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":11.36334,"latencyP50":37.8969,"badge":"verified","holdoutGap":0.022482,"holdoutLimit":0.1066,"topicIdx":{"politics":-11.4,"economy":40.1,"technology":17.4,"science-health":-11.3,"identity":-31.2,"society":84.2,"global":-2.9},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--claude-sonnet-5.5--tdefault--tplcurrent--bfbdc53f"},"gss":{"sps":0.761179,"idx":9.8,"lo":-2.5,"hi":22.3,"pDist":0.639754,"pRank":0.676034,"pRefuse":0.967748,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":10.86526,"latencyP50":25.4797,"badge":"verified","holdoutGap":0.011028,"holdoutLimit":0.129099,"topicIdx":{"politics":-4.8,"economy":10.6,"technology":33.9,"science-health":33.4,"identity":7.8,"society":16.4,"global":8.8},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"althing--claude-sonnet-5.5--tdefault--tplcurrent--7f8d308c"},"opinionsqa":{"sps":0.815734,"idx":22.3,"lo":13.6,"hi":30.9,"pDist":0.691748,"pRank":0.761181,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":8.265435,"latencyP50":31.139,"badge":"verified","holdoutGap":0.019514,"holdoutLimit":0.104257,"topicIdx":{"politics":-19.8,"economy":-11,"technology":26.8,"science-health":24.1,"identity":4,"society":31.5},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"althing--claude-sonnet-5.5--tdefault--tplcurrent--be6f8875"},"subpop":{"sps":0.722721,"idx":-14.6,"lo":-30.1,"hi":-0.6,"pDist":0.555496,"pRank":0.63278,"pRefuse":0.979886,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":10.595235,"latencyP50":42.9515,"badge":"verified","holdoutGap":0.00751,"holdoutLimit":0.094491,"topicIdx":{"politics":-90.9,"economy":29.2,"technology":20.9,"science-health":18.9,"identity":-14.9,"society":12.3,"global":3.9},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--claude-sonnet-5.5--tdefault--tplcurrent--1b3897e6"}},"coverage":4,"index":3.4,"indexLo":-2.9,"indexHi":9.7,"cost100":10.2723,"latencyP50":34.37,"flagged":0,"runs":4},{"key":"raw|google/gemini-2.5-flash|default|current|||","base":"Gemini 2.5 Flash","name":"Gemini 2.5 Flash, raw","short":"Gemini 2.5 Flash","baseId":"google/gemini-2.5-flash","members":[],"lab":"Google","method":"raw","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.765301,"idx":-3.9,"lo":-17.2,"hi":9.9,"pDist":0.673689,"pRank":0.64248,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.107343,"latencyP50":1.797,"badge":"verified","holdoutGap":0.081268,"holdoutLimit":0.1066,"topicIdx":{"politics":-9.4,"economy":62,"technology":25.7,"science-health":12.7,"identity":-11.1,"society":56.7,"global":-15.4},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"raw--gemini-2.5-flash--tdefault--tplcurrent--07d79295"},"gss":{"sps":0.789471,"idx":20.5,"lo":8.9,"hi":31.1,"pDist":0.692354,"pRank":0.704008,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":0.106955,"latencyP50":1.625,"badge":"verified","holdoutGap":0.026681,"holdoutLimit":0.129099,"topicIdx":{"politics":43.5,"economy":-22,"technology":21.9,"science-health":16.9,"identity":2.7,"society":25.7,"global":0.4},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"raw--gemini-2.5-flash--tdefault--tplcurrent--2d09bedc"},"opinionsqa":{"sps":0.82737,"idx":27.3,"lo":19.6,"hi":34.7,"pDist":0.729069,"pRank":0.759068,"pRefuse":0.993972,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.099174,"latencyP50":1.703,"badge":"verified","holdoutGap":0.030268,"holdoutLimit":0.104257,"topicIdx":{"politics":63.5,"economy":-2.6,"technology":27.6,"science-health":5.9,"identity":27.4,"society":38.1},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"raw--gemini-2.5-flash--tdefault--tplcurrent--2f651edf"},"subpop":{"sps":0.671725,"idx":-36,"lo":-42.6,"hi":-29.8,"pDist":0.667429,"pRank":0.683949,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":15,"cost100":0.050402,"latencyP50":2.313,"badge":"verified","holdoutGap":0.052033,"holdoutLimit":0.094491,"topicIdx":{"politics":-53.4,"economy":-13.4,"technology":-30.4,"science-health":-53.3,"identity":-28.8,"society":-21.9,"global":-45.5},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"raw--gemini-2.5-flash--tdefault--tplcurrent--4a7db0f2"}},"coverage":4,"index":2,"indexLo":-3,"indexHi":7,"cost100":0.091,"latencyP50":1.86,"flagged":0,"runs":4},{"key":"raw|anthropic/claude-sonnet-4-6|default|current|||","base":"Claude Sonnet 4.6","name":"Claude Sonnet 4.6, raw","short":"Sonnet 4.6","baseId":"anthropic/claude-sonnet-4-6","members":[],"lab":"Anthropic","method":"raw","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.728362,"idx":-20.5,"lo":-37,"hi":-4.1,"pDist":0.570611,"pRank":0.634743,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":1.30707,"latencyP50":3.938,"badge":"verified","holdoutGap":0.011592,"holdoutLimit":0.1066,"topicIdx":{"politics":-52.5,"economy":5.6,"technology":12.2,"science-health":-17.8,"identity":-45.9,"society":56.7,"global":-1.9},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"raw--claude-sonnet-4.6--tdefault--tplcurrent--c982ba2c"},"gss":{"sps":0.774309,"idx":14.8,"lo":1.5,"hi":27,"pDist":0.631852,"pRank":0.719026,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":1.25988,"latencyP50":4.312,"badge":"verified","holdoutGap":0.043448,"holdoutLimit":0.129099,"topicIdx":{"politics":28.8,"economy":-8.7,"technology":47.9,"science-health":17,"identity":-1.6,"society":6.1,"global":21.6},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"raw--claude-sonnet-4.6--tdefault--tplcurrent--3c0b8591"},"opinionsqa":{"sps":0.807934,"idx":19,"lo":10.4,"hi":27.7,"pDist":0.675926,"pRank":0.753603,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":1.18752,"latencyP50":4.906,"badge":"verified","holdoutGap":0.045606,"holdoutLimit":0.104257,"topicIdx":{"politics":-73.4,"economy":-32.7,"technology":15,"science-health":19.4,"identity":28.7,"society":31.6},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"raw--claude-sonnet-4.6--tdefault--tplcurrent--2e68d4a6"},"subpop":{"sps":0.74291,"idx":-6.1,"lo":-18.6,"hi":5.2,"pDist":0.571304,"pRank":0.677497,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":1.23408,"latencyP50":3.907,"badge":"verified","holdoutGap":0.042517,"holdoutLimit":0.094491,"topicIdx":{"politics":-28.9,"economy":-10.7,"technology":10.9,"science-health":7.7,"identity":-8.1,"society":-2.6,"global":-7.9},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"raw--claude-sonnet-4.6--tdefault--tplcurrent--726d62e9"}},"coverage":4,"index":1.8,"indexLo":-4.6,"indexHi":8.2,"cost100":1.2471,"latencyP50":4.27,"flagged":0,"runs":4},{"key":"althing|google/gemini-2.5-flash-lite|default|current|||","base":"Gemini 2.5 Flash-Lite","name":"Gemini 2.5 Flash-Lite, persona","short":"Gemini 2.5 Flash-Lite","baseId":"google/gemini-2.5-flash-lite","members":[],"lab":"Google","method":"althing","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.771426,"idx":-1.2,"lo":-15,"hi":11.4,"pDist":0.681164,"pRank":0.65338,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.04547,"latencyP50":14.5219,"badge":"verified","holdoutGap":0.084841,"holdoutLimit":0.1066,"topicIdx":{"politics":-4.1,"economy":36.5,"technology":34,"science-health":25.6,"identity":-40.1,"society":84.2,"global":-6.2},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"althing--gemini-2.5-flash-lite--tdefault--tplcurrent--3ce0d15e"},"gss":{"sps":0.818377,"idx":31.5,"lo":21.5,"hi":41,"pDist":0.71684,"pRank":0.76624,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":0.044057,"latencyP50":9.1344,"badge":"verified","holdoutGap":0.004679,"holdoutLimit":0.129099,"topicIdx":{"politics":52.6,"economy":2.4,"technology":44.9,"science-health":46.7,"identity":11.6,"society":26.6,"global":-24.4},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"althing--gemini-2.5-flash-lite--tdefault--tplcurrent--ea8921f1"},"subpop":{"sps":0.682187,"idx":-31.6,"lo":-36.8,"hi":-27,"pDist":0.701199,"pRank":0.741179,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":15,"cost100":0.023044,"latencyP50":6.8235,"badge":"verified","holdoutGap":0.010693,"holdoutLimit":0.094491,"topicIdx":{"politics":-33.7,"economy":-43.2,"technology":-31,"science-health":-20.6,"identity":-27.4,"society":-31.4,"global":-36},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--gemini-2.5-flash-lite--tdefault--tplcurrent--2dd7b214"}},"coverage":3,"index":-0.4,"indexLo":-6.1,"indexHi":5.3,"cost100":0.0375,"latencyP50":10.16,"flagged":0,"runs":3},{"key":"raw|openai/gpt-4o-mini|default|current|||","base":"GPT-4o mini","name":"GPT-4o mini, raw","short":"GPT-4o mini","baseId":"openai/gpt-4o-mini","members":[],"lab":"OpenAI","method":"raw","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.750386,"idx":-10.6,"lo":-25.5,"hi":2.8,"pDist":0.63811,"pRank":0.633314,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.05302,"latencyP50":6.437,"badge":"verified","holdoutGap":0.019672,"holdoutLimit":0.1066,"topicIdx":{"politics":-17.6,"economy":22.6,"technology":30.9,"science-health":42.9,"identity":-22.4,"society":56.7,"global":-21.5},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"raw--gpt-4o-mini--tdefault--tplcurrent--70d804f8"},"gss":{"sps":0.804234,"idx":26.1,"lo":16.6,"hi":35.5,"pDist":0.685544,"pRank":0.755108,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":0.051289,"latencyP50":6.219,"badge":"verified","holdoutGap":0.005802,"holdoutLimit":0.129099,"topicIdx":{"politics":42.3,"economy":-3.7,"technology":23,"science-health":48,"identity":6.9,"society":29.1,"global":-8.1},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"raw--gpt-4o-mini--tdefault--tplcurrent--d46362fb"},"opinionsqa":{"sps":0.816512,"idx":22.7,"lo":13.1,"hi":31.4,"pDist":0.702388,"pRank":0.752875,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.048483,"latencyP50":6.235,"badge":"verified","holdoutGap":0.023523,"holdoutLimit":0.104257,"topicIdx":{"politics":64.7,"economy":-0.2,"technology":13.5,"science-health":25,"identity":17.8,"society":42},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"raw--gpt-4o-mini--tdefault--tplcurrent--2b9ea7fa"},"subpop":{"sps":0.657987,"idx":-41.8,"lo":-48.6,"hi":-35.5,"pDist":0.625533,"pRank":0.696365,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":15,"cost100":0.025003,"latencyP50":3.203,"badge":"verified","holdoutGap":0.073541,"holdoutLimit":0.094491,"topicIdx":{"politics":-40.8,"economy":-10.4,"technology":-48.6,"science-health":-38.8,"identity":-43.4,"society":-40.8,"global":-33.1},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"raw--gpt-4o-mini--tdefault--tplcurrent--4dba4f5c"}},"coverage":4,"index":-0.9,"indexLo":-6,"indexHi":4.2,"cost100":0.0444,"latencyP50":5.52,"flagged":0,"runs":4},{"key":"raw|anthropic/claude-sonnet-5-5|default|current|||","base":"Claude Sonnet 5.5","name":"Claude Sonnet 5.5, raw","short":"Sonnet 5.5","baseId":"anthropic/claude-sonnet-5-5","members":[],"lab":"Anthropic","method":"raw","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.755209,"idx":-8.4,"lo":-23.4,"hi":5.7,"pDist":0.597967,"pRank":0.690069,"pRefuse":0.97759,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":1.64214,"latencyP50":4.078,"badge":"verified","holdoutGap":0.012162,"holdoutLimit":0.1066,"topicIdx":{"politics":-26.7,"economy":34.9,"technology":19.5,"science-health":13.2,"identity":-33.4,"society":56.7,"global":-3.9},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"raw--claude-sonnet-5.5--tdefault--tplcurrent--028309be"},"gss":{"sps":0.735957,"idx":0.2,"lo":-15.8,"hi":14,"pDist":0.581025,"pRank":0.654795,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":1.50396,"latencyP50":4.25,"badge":"verified","holdoutGap":0.098547,"holdoutLimit":0.129099,"topicIdx":{"politics":9.6,"economy":-11.4,"technology":26.7,"science-health":8.6,"identity":-8.9,"society":-3.8,"global":-71.6},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"raw--claude-sonnet-5.5--tdefault--tplcurrent--70733fb2"},"opinionsqa":{"sps":0.78889,"idx":11,"lo":-0.6,"hi":22.1,"pDist":0.653294,"pRank":0.725647,"pRefuse":0.987728,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":1.51677,"latencyP50":4.062,"badge":"verified","holdoutGap":0.029147,"holdoutLimit":0.104257,"topicIdx":{"politics":-43.6,"economy":-86.9,"technology":17,"science-health":22.5,"identity":-17.4,"society":30},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"raw--claude-sonnet-5.5--tdefault--tplcurrent--75ca2916"},"subpop":{"sps":0.736138,"idx":-8.9,"lo":-24.4,"hi":3.8,"pDist":0.568987,"pRank":0.672073,"pRefuse":0.967355,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":1.647465,"latencyP50":4.234,"badge":"verified","holdoutGap":0.019019,"holdoutLimit":0.094491,"topicIdx":{"politics":-41.7,"economy":-10.7,"technology":7.9,"science-health":5.4,"identity":-25.7,"society":18.2,"global":7.9},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"raw--claude-sonnet-5.5--tdefault--tplcurrent--358bacff"}},"coverage":4,"index":-1.5,"indexLo":-8.4,"indexHi":5.4,"cost100":1.5776,"latencyP50":4.16,"flagged":0,"runs":4},{"key":"raw|openai/gpt-6-luna|default|current|||codex","base":"GPT-6 Luna","name":"GPT-6 Luna via codex, raw","short":"GPT-6 Luna via codex","baseId":"openai/gpt-6-luna","members":[],"lab":"OpenAI","method":"raw","variant":null,"harness":"codex","ds":{"globalopinionqa":{"sps":0.760168,"idx":-6.2,"lo":-22.4,"hi":8.4,"pDist":0.673935,"pRank":0.626836,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":29.781,"badge":"verified","holdoutGap":0.039023,"holdoutLimit":0.1066,"topicIdx":{"politics":-37.9,"economy":26.6,"technology":28.3,"science-health":4.3,"identity":-15.8,"society":56.7,"global":6.9},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"raw--gpt-6-luna--tdefault--tplcurrent--cf7c2476"},"gss":{"sps":0.710286,"idx":-9.5,"lo":-22.5,"hi":3.7,"pDist":0.587964,"pRank":0.570845,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":29.578,"badge":"verified","holdoutGap":0.055376,"holdoutLimit":0.129099,"topicIdx":{"politics":15.8,"economy":-70,"technology":-9.6,"science-health":-5.9,"identity":-19.2,"society":-26,"global":13.8},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"raw--gpt-6-luna--tdefault--tplcurrent--e085f95b"},"opinionsqa":{"sps":0.72701,"idx":-15.1,"lo":-29.2,"hi":-1.5,"pDist":0.584309,"pRank":0.602448,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":28.937,"badge":"flagged","holdoutGap":0.20111,"holdoutLimit":0.104257,"topicIdx":{"politics":-56,"economy":-80,"technology":31.2,"science-health":-52.7,"identity":-21.5,"society":-64},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"raw--gpt-6-luna--tdefault--tplcurrent--98c3512d"},"subpop":{"sps":0.742985,"idx":-6.1,"lo":-19.9,"hi":7.3,"pDist":0.62384,"pRank":0.625185,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":29.532,"badge":"verified","holdoutGap":0.016936,"holdoutLimit":0.094491,"topicIdx":{"politics":-21.2,"economy":-74.6,"technology":24.7,"science-health":-33.6,"identity":-1.2,"society":-36.3,"global":16.7},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"raw--gpt-6-luna--tdefault--tplcurrent--5c43d9eb"}},"coverage":4,"index":-9.2,"indexLo":-16.2,"indexHi":-2.2,"cost100":null,"latencyP50":29.46,"flagged":1,"runs":4},{"key":"raw|openai/gpt-6-luna|default|current|||","base":"GPT-6 Luna","name":"GPT-6 Luna, raw","short":"GPT-6 Luna","baseId":"openai/gpt-6-luna","members":[],"lab":"OpenAI","method":"raw","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.741805,"idx":-14.4,"lo":-31.3,"hi":1.2,"pDist":0.626637,"pRank":0.619722,"pRefuse":0.979055,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":8.657,"badge":"verified","holdoutGap":0.046192,"holdoutLimit":0.1066,"topicIdx":{"politics":-48.9,"economy":49.7,"technology":17.6,"science-health":1.3,"identity":-27,"society":56.7,"global":-3.6},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"raw--gpt-6-luna--tdefault--tplcurrent--7a45dae5"},"gss":{"sps":0.724092,"idx":-4.3,"lo":-19.5,"hi":10.8,"pDist":0.585617,"pRank":0.614608,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":9.125,"badge":"verified","holdoutGap":0.107089,"holdoutLimit":0.129099,"topicIdx":{"politics":13.3,"economy":-85.1,"technology":19.5,"science-health":15.3,"identity":-9.1,"society":-26.7,"global":29.3},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"raw--gpt-6-luna--tdefault--tplcurrent--d8ed4aa2"},"opinionsqa":{"sps":0.714129,"idx":-20.6,"lo":-35.2,"hi":-6,"pDist":0.555841,"pRank":0.592276,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":8.719,"badge":"flagged","holdoutGap":0.15137,"holdoutLimit":0.104257,"topicIdx":{"politics":-63.7,"economy":-59.6,"technology":18.9,"science-health":-32.7,"identity":-22.6,"society":-73.8},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"raw--gpt-6-luna--tdefault--tplcurrent--f4299b61"},"subpop":{"sps":0.723542,"idx":-14.2,"lo":-28.2,"hi":-0.5,"pDist":0.572214,"pRank":0.61848,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":9.766,"badge":"verified","holdoutGap":0.049473,"holdoutLimit":0.094491,"topicIdx":{"politics":-22.2,"economy":-89.4,"technology":14,"science-health":-69.2,"identity":-6.9,"society":-44,"global":7.4},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"raw--gpt-6-luna--tdefault--tplcurrent--3b7430b9"}},"coverage":4,"index":-13.4,"indexLo":-20.9,"indexHi":-5.9,"cost100":null,"latencyP50":9.07,"flagged":1,"runs":4},{"key":"althing|anthropic/claude-haiku-4-5|0.85|current|||","base":"Claude Haiku 4.5","name":"Claude Haiku 4.5, persona (temperature 0.85)","short":"Haiku 4.5","baseId":"anthropic/claude-haiku-4-5","members":[],"lab":"Anthropic","method":"althing","variant":"temperature 0.85","harness":null,"ds":{"subpop":{"sps":0.725414,"idx":-13.4,"lo":-24.3,"hi":-1.9,"pDist":0.656265,"pRank":0.628435,"pRefuse":0.891541,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.432,"latencyP50":24.2062,"badge":"verified","holdoutGap":0.013728,"holdoutLimit":0.094491,"topicIdx":{"politics":-47.5,"economy":-5.7,"technology":18.2,"science-health":11,"identity":-29.3,"society":-13.2,"global":-16.7},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--claude-haiku-4.5--t0.85--tplcurrent--b3a09624"}},"coverage":1,"index":-13.4,"indexLo":-24.6,"indexHi":-2.2,"cost100":0.432,"latencyP50":24.21,"flagged":0,"runs":1},{"key":"raw|anthropic/claude-haiku-4-5|default|current|||","base":"Claude Haiku 4.5","name":"Claude Haiku 4.5, raw","short":"Haiku 4.5","baseId":"anthropic/claude-haiku-4-5","members":[],"lab":"Anthropic","method":"raw","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.669765,"idx":-46.7,"lo":-63.4,"hi":-29,"pDist":0.666815,"pRank":0.625964,"pRefuse":0.716516,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.673935,"latencyP50":2.094,"badge":"flagged","holdoutGap":0.142521,"holdoutLimit":0.1066,"topicIdx":{"politics":-65.8,"economy":-1.3,"technology":24.7,"science-health":7.4,"identity":-58.7,"society":56.7,"global":-55.6},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"raw--claude-haiku-4.5--tdefault--tplcurrent--79c246ae"},"gss":{"sps":0.734393,"idx":-0.4,"lo":-14.4,"hi":14.7,"pDist":0.607091,"pRank":0.677266,"pRefuse":0.918822,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":0.476947,"latencyP50":2.078,"badge":"verified","holdoutGap":0.064564,"holdoutLimit":0.129099,"topicIdx":{"politics":16.2,"economy":-51.8,"technology":15.6,"science-health":-22,"identity":-9.6,"society":10,"global":0.4},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"raw--claude-haiku-4.5--tdefault--tplcurrent--aaec9e4f"},"opinionsqa":{"sps":0.765404,"idx":1.1,"lo":-11.4,"hi":13.2,"pDist":0.670482,"pRank":0.720055,"pRefuse":0.905674,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.479635,"latencyP50":2.032,"badge":"verified","holdoutGap":0.049577,"holdoutLimit":0.104257,"topicIdx":{"politics":-97.3,"economy":-37.2,"technology":17.4,"science-health":-46,"identity":-2.4,"society":3.3},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"raw--claude-haiku-4.5--tdefault--tplcurrent--184822f4"},"subpop":{"sps":0.706013,"idx":-21.6,"lo":-37.3,"hi":-7.7,"pDist":0.607038,"pRank":0.680168,"pRefuse":0.830832,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.549935,"latencyP50":2.032,"badge":"verified","holdoutGap":0.063629,"holdoutLimit":0.094491,"topicIdx":{"politics":-59.3,"economy":-91.4,"technology":6.7,"science-health":-38.4,"identity":-9.2,"society":5.6,"global":-43.8},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"raw--claude-haiku-4.5--tdefault--tplcurrent--58a8b658"}},"coverage":4,"index":-16.9,"indexLo":-24.3,"indexHi":-9.5,"cost100":0.5451,"latencyP50":2.06,"flagged":1,"runs":4},{"key":"raw|anthropic/claude-haiku-5-5|default|current|||","base":"Claude Haiku 5.5","name":"Claude Haiku 5.5, raw","short":"Haiku 5.5","baseId":"anthropic/claude-haiku-5-5","members":[],"lab":"Anthropic","method":"raw","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.656076,"idx":-52.8,"lo":-71.5,"hi":-33.8,"pDist":0.609942,"pRank":0.581311,"pRefuse":0.776976,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.705105,"latencyP50":3.562,"badge":"verified","holdoutGap":0.073819,"holdoutLimit":0.1066,"topicIdx":{"politics":-96.9,"economy":-19.8,"technology":17.6,"science-health":8.7,"identity":-20.6,"society":56.7,"global":-52},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"raw--claude-haiku-5.5--tdefault--tplcurrent--d3148d0d"},"gss":{"sps":0.730279,"idx":-1.9,"lo":-17.6,"hi":12.7,"pDist":0.598136,"pRank":0.644118,"pRefuse":0.948583,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":0.522267,"latencyP50":3.407,"badge":"verified","holdoutGap":0.034421,"holdoutLimit":0.129099,"topicIdx":{"politics":18.4,"economy":-54.2,"technology":9.6,"science-health":26.5,"identity":-15.9,"society":-12.7,"global":-60},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"raw--claude-haiku-5.5--tdefault--tplcurrent--ad83473b"},"opinionsqa":{"sps":0.712655,"idx":-21.2,"lo":-39.5,"hi":-4.6,"pDist":0.651441,"pRank":0.68929,"pRefuse":0.797235,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.647395,"latencyP50":3.438,"badge":"verified","holdoutGap":0.045832,"holdoutLimit":0.104257,"topicIdx":{"politics":66,"economy":-17.3,"technology":18.1,"science-health":-113.7,"identity":-53.7,"society":-37.3},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"raw--claude-haiku-5.5--tdefault--tplcurrent--1e56c197"},"subpop":{"sps":0.695092,"idx":-26.2,"lo":-44.7,"hi":-7.7,"pDist":0.570752,"pRank":0.653458,"pRefuse":0.861065,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.60126,"latencyP50":3.421,"badge":"verified","holdoutGap":0.021314,"holdoutLimit":0.094491,"topicIdx":{"politics":-36.3,"economy":-98.9,"technology":10,"science-health":-4.4,"identity":-49.3,"society":-75.1,"global":26.5},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"raw--claude-haiku-5.5--tdefault--tplcurrent--ee69632d"}},"coverage":4,"index":-25.5,"indexLo":-34.3,"indexHi":-16.7,"cost100":0.619,"latencyP50":3.46,"flagged":0,"runs":4},{"key":"raw|anthropic/claude-sonnet-4|default|current|||","base":"Claude Sonnet 4","name":"Claude Sonnet 4, raw","short":"Sonnet 4","baseId":"anthropic/claude-sonnet-4","members":[],"lab":"Anthropic","method":"raw","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.631055,"idx":-64.1,"lo":-82.3,"hi":-45.5,"pDist":0.688619,"pRank":0.595105,"pRefuse":0.609442,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":2.07228,"latencyP50":6.531,"badge":"verified","holdoutGap":0.092658,"holdoutLimit":0.1066,"topicIdx":{"politics":-95.9,"economy":7.9,"technology":17.6,"science-health":-10.1,"identity":-98.8,"society":56.7,"global":-64.4},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"raw--claude-sonnet-4--tdefault--tplcurrent--d4b3dbc0"},"gss":{"sps":0.720003,"idx":-5.8,"lo":-21.1,"hi":10,"pDist":0.610685,"pRank":0.651363,"pRefuse":0.897959,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":1.40864,"latencyP50":5.375,"badge":"verified","holdoutGap":0.053542,"holdoutLimit":0.129099,"topicIdx":{"politics":-2.3,"economy":-84.9,"technology":28.8,"science-health":16.4,"identity":-18.8,"society":13.4,"global":-18.7},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"raw--claude-sonnet-4--tdefault--tplcurrent--6cb65d42"},"opinionsqa":{"sps":0.719758,"idx":-18.2,"lo":-35.9,"hi":-0.4,"pDist":0.618368,"pRank":0.665165,"pRefuse":0.875741,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":1.352805,"latencyP50":4.875,"badge":"flagged","holdoutGap":0.104554,"holdoutLimit":0.104257,"topicIdx":{"politics":-194.9,"economy":-138.5,"technology":28,"science-health":-163.2,"identity":-0.6,"society":-25.5},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"raw--claude-sonnet-4--tdefault--tplcurrent--09b96054"},"subpop":{"sps":0.66542,"idx":-38.7,"lo":-58.9,"hi":-20.6,"pDist":0.549295,"pRank":0.613825,"pRefuse":0.833139,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":1.50759,"latencyP50":4.938,"badge":"verified","holdoutGap":0.056701,"holdoutLimit":0.094491,"topicIdx":{"politics":-123.2,"economy":17.9,"technology":16.9,"science-health":-64.4,"identity":-3.9,"society":-42,"global":-67.9},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"raw--claude-sonnet-4--tdefault--tplcurrent--62d154a0"}},"coverage":4,"index":-31.7,"indexLo":-40.6,"indexHi":-22.8,"cost100":1.5853,"latencyP50":5.43,"flagged":1,"runs":4},{"key":"althing|anthropic/claude-haiku-4-5|0.85|demo|||","base":"Claude Haiku 4.5","name":"Claude Haiku 4.5, persona (temperature 0.85, demo template)","short":"Haiku 4.5","baseId":"anthropic/claude-haiku-4-5","members":[],"lab":"Anthropic","method":"althing","variant":"temperature 0.85, demo template","harness":null,"ds":{"subpop":{"sps":0.667463,"idx":-37.8,"lo":-50.8,"hi":-24.3,"pDist":0.666782,"pRank":0.637453,"pRefuse":0.698153,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.432,"latencyP50":25.0594,"badge":"verified","holdoutGap":0.026658,"holdoutLimit":0.094491,"topicIdx":{"politics":-66.8,"economy":6.8,"technology":-9.9,"science-health":-11.8,"identity":-76,"society":-18.7,"global":-54.5},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--claude-haiku-4.5--t0.85--tpldemo--440fd214"}},"coverage":1,"index":-37.8,"indexLo":-51,"indexHi":-24.5,"cost100":0.432,"latencyP50":25.06,"flagged":0,"runs":1},{"key":"althing|anthropic/claude-haiku-4-5|0.85|minimal|||","base":"Claude Haiku 4.5","name":"Claude Haiku 4.5, persona (temperature 0.85, minimal template)","short":"Haiku 4.5","baseId":"anthropic/claude-haiku-4-5","members":[],"lab":"Anthropic","method":"althing","variant":"temperature 0.85, minimal template","harness":null,"ds":{"subpop":{"sps":0.648169,"idx":-45.9,"lo":-61.2,"hi":-30.2,"pDist":0.639114,"pRank":0.603754,"pRefuse":0.70164,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.432,"latencyP50":24.3375,"badge":"verified","holdoutGap":0.063326,"holdoutLimit":0.094491,"topicIdx":{"politics":-74,"economy":-40.9,"technology":11.3,"science-health":-17.3,"identity":-87.6,"society":-39.2,"global":-142},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--claude-haiku-4.5--t0.85--tplminimal--592ecfd1"}},"coverage":1,"index":-45.9,"indexLo":-61.4,"indexHi":-30.4,"cost100":0.432,"latencyP50":24.34,"flagged":0,"runs":1},{"key":"althing|anthropic/claude-haiku-4-5|0.85|values|||","base":"Claude Haiku 4.5","name":"Claude Haiku 4.5, persona (temperature 0.85, values template)","short":"Haiku 4.5","baseId":"anthropic/claude-haiku-4-5","members":[],"lab":"Anthropic","method":"althing","variant":"temperature 0.85, values template","harness":null,"ds":{"subpop":{"sps":0.637425,"idx":-50.4,"lo":-66.4,"hi":-35.9,"pDist":0.632338,"pRank":0.60961,"pRefuse":0.670325,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0.432,"latencyP50":24.2766,"badge":"verified","holdoutGap":0.08954,"holdoutLimit":0.094491,"topicIdx":{"politics":-92.2,"economy":-17,"technology":-5.9,"science-health":-24.5,"identity":-83.3,"society":-29.9,"global":-102.9},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"althing--claude-haiku-4.5--t0.85--tplvalues--429cc601"}},"coverage":1,"index":-50.4,"indexLo":-65.6,"indexHi":-35.1,"cost100":0.432,"latencyP50":24.28,"flagged":0,"runs":1},{"key":"raw|google/gemini-3.8-flash|default|current|||","base":"Gemini 3.8 Flash","name":"Gemini 3.8 Flash, raw","short":"Gemini 3.8 Flash","baseId":"google/gemini-3.8-flash","members":[],"lab":"Google","method":"raw","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.694283,"idx":-35.7,"lo":-54.4,"hi":-18.5,"pDist":0.562682,"pRank":0.58987,"pRefuse":0.930296,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":2.29254,"latencyP50":16.844,"badge":"verified","holdoutGap":0.022873,"holdoutLimit":0.1066,"topicIdx":{"politics":-62.5,"economy":18,"technology":30.6,"science-health":-35.8,"identity":-53.2,"society":56.7,"global":-31.6},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"raw--gemini-3.8-flash--tdefault--tplcurrent--8f94fc85"},"gss":{"sps":0.53108,"idx":-77.6,"lo":-90.5,"hi":-63.2,"pDist":0.290304,"pRank":0.330886,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":1.457996,"latencyP50":14.281,"badge":"verified","holdoutGap":0.066033,"holdoutLimit":0.129099,"topicIdx":{"politics":-85.5,"economy":-105.9,"technology":-52.2,"science-health":-87.1,"identity":-60.5,"society":-73.2,"global":-125.2},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"raw--gemini-3.8-flash--tdefault--tplcurrent--b56a1300"},"opinionsqa":{"sps":0.701311,"idx":-26,"lo":-41.2,"hi":-9.9,"pDist":0.538692,"pRank":0.570969,"pRefuse":0.994272,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":1.55559,"latencyP50":15.984,"badge":"verified","holdoutGap":0.024264,"holdoutLimit":0.104257,"topicIdx":{"politics":-73.4,"economy":-107.7,"technology":7.6,"science-health":-107.5,"identity":-83.9,"society":-3.7},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"raw--gemini-3.8-flash--tdefault--tplcurrent--cb4842bd"},"subpop":{"sps":0.599975,"idx":-66.2,"lo":-83.5,"hi":-49.7,"pDist":0.379499,"pRank":0.440494,"pRefuse":0.979931,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":1.607282,"latencyP50":15.062,"badge":"verified","holdoutGap":0.070071,"holdoutLimit":0.094491,"topicIdx":{"politics":-135.2,"economy":-74.8,"technology":11,"science-health":-115.8,"identity":-108.7,"society":-27.5,"global":-114.4},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"raw--gemini-3.8-flash--tdefault--tplcurrent--37e38f3f"}},"coverage":4,"index":-51.4,"indexLo":-59.5,"indexHi":-43.3,"cost100":1.7284,"latencyP50":15.54,"flagged":0,"runs":4},{"key":"raw|meta/muse-glimmer-30b|default|current|||","base":"Muse Glimmer 30B","name":"Muse Glimmer 30B, raw","short":"Muse Glimmer 30B","baseId":"meta/muse-glimmer-30b","members":[],"lab":"Meta","method":"raw","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.655351,"idx":-53.2,"lo":-70.7,"hi":-36.6,"pDist":0.564766,"pRank":0.537185,"pRefuse":0.864102,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":7.125,"badge":"verified","holdoutGap":0.013502,"holdoutLimit":0.1066,"topicIdx":{"politics":-84.9,"economy":22.2,"technology":-7.4,"science-health":-24.4,"identity":-77.3,"society":78.4,"global":-48.4},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"raw--muse-glimmer-30b--tdefault--tplcurrent--e1e6cae4"},"gss":{"sps":0.543755,"idx":-72.7,"lo":-84.3,"hi":-61.2,"pDist":0.319182,"pRank":0.341311,"pRefuse":0.970771,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":5.688,"badge":"verified","holdoutGap":0.002567,"holdoutLimit":0.129099,"topicIdx":{"politics":-59.6,"economy":-98.3,"technology":-84.9,"science-health":-84.2,"identity":-70.4,"society":-72.3,"global":-115.4},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"raw--muse-glimmer-30b--tdefault--tplcurrent--6a7d52ac"},"opinionsqa":{"sps":0.689036,"idx":-31.2,"lo":-44.2,"hi":-17.5,"pDist":0.576881,"pRank":0.521183,"pRefuse":0.969043,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":6.235,"badge":"verified","holdoutGap":0.097584,"holdoutLimit":0.104257,"topicIdx":{"politics":-54.7,"economy":-54.1,"technology":2.8,"science-health":-88.6,"identity":-70,"society":-39.9},"topicN":{"politics":1,"economy":4,"technology":44,"science-health":10,"identity":15,"society":26},"configId":"raw--muse-glimmer-30b--tdefault--tplcurrent--5b4d5032"},"subpop":{"sps":0.632127,"idx":-52.6,"lo":-67.4,"hi":-38.8,"pDist":0.463871,"pRank":0.467428,"pRefuse":0.965081,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":null,"latencyP50":6.718,"badge":"verified","holdoutGap":0.047217,"holdoutLimit":0.094491,"topicIdx":{"politics":-95.5,"economy":-93.5,"technology":-3.7,"science-health":-61.4,"identity":-58.7,"society":-42,"global":-96.4},"topicN":{"politics":25,"economy":4,"technology":31,"science-health":4,"identity":15,"society":14,"global":7},"configId":"raw--muse-glimmer-30b--tdefault--tplcurrent--029f8248"}},"coverage":4,"index":-52.4,"indexLo":-59.5,"indexHi":-45.3,"cost100":null,"latencyP50":6.44,"flagged":0,"runs":4}],"references":{"random":{"key":"baseline|random-baseline|default|current|||","base":"Random answers","name":"Random answers","short":"Random answers","baseId":"random-baseline","members":[],"lab":"Other","method":"baseline","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.774017,"idx":0,"lo":-10.4,"hi":10.1,"pDist":0.890422,"pRank":0.451897,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0,"latencyP50":0.0001,"badge":"verified","holdoutGap":0.02962,"holdoutLimit":0.1066,"topicIdx":{"politics":0.6,"economy":-24,"technology":-23.7,"science-health":-28.2,"identity":14.5,"society":-4.9,"global":6.8},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"baseline--random-baseline--tdefault--tplcurrent--8855add7"},"gss":{"sps":0.73538,"idx":0,"lo":-7.5,"hi":7.7,"pDist":0.816904,"pRank":0.417184,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":0,"latencyP50":0.0001,"badge":"verified","holdoutGap":0.06313,"holdoutLimit":0.129099,"topicIdx":{"politics":-1,"economy":-17.7,"technology":28.6,"science-health":-3.6,"identity":12.4,"society":-16.6,"global":-18.6},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"baseline--random-baseline--tdefault--tplcurrent--2ad2dd26"},"ntia":{"sps":0.788153,"idx":0,"lo":-17.1,"hi":16.2,"pDist":0.889776,"pRank":0.474684,"pRefuse":1,"n":79,"runs":1,"samplesPerQuestion":30,"cost100":0,"latencyP50":0,"badge":"verified","holdoutGap":0.047754,"holdoutLimit":0.089803,"topicIdx":{"economy":-30.4,"technology":6},"topicN":{"economy":4,"technology":66},"configId":"baseline--random-baseline--tdefault--tplcurrent--7ba7f86c"},"opinionsqa":{"sps":0.762852,"idx":0,"lo":-3.3,"hi":3.3,"pDist":0.805999,"pRank":0.49303,"pRefuse":0.989527,"n":684,"runs":15,"samplesPerQuestion":30,"cost100":0,"latencyP50":null,"badge":"verified","holdoutGap":0.019515,"holdoutLimit":0.05,"topicIdx":{"politics":-11,"economy":2.7,"technology":5.7,"science-health":0.7,"identity":-3.6,"society":-2.1,"global":7.1},"topicN":{"politics":80,"economy":73,"technology":146,"science-health":140,"identity":116,"society":82,"global":47},"configId":"baseline--random-baseline--tdefault--tplcurrent--971f8ad1"},"subpop":{"sps":0.757397,"idx":0,"lo":-5.8,"hi":5.7,"pDist":0.816139,"pRank":0.480518,"pRefuse":0.975532,"n":200,"runs":1,"samplesPerQuestion":30,"cost100":0,"latencyP50":null,"badge":"verified","holdoutGap":0.007889,"holdoutLimit":0.068041,"topicIdx":{"politics":15.1,"economy":-16.1,"technology":5.6,"science-health":5.9,"identity":-14.5,"society":-5.5,"global":-2.6},"topicN":{"politics":42,"economy":5,"technology":65,"science-health":4,"identity":56,"society":18,"global":10},"configId":"baseline--random-baseline--tdefault--tplcurrent--c53167f6"}},"coverage":4,"index":0,"indexLo":-3.6,"indexHi":3.6,"cost100":0,"latencyP50":0,"flagged":0,"runs":19},"majority":{"key":"baseline|majority-baseline|default|current|||","base":"Majority answer","name":"Majority answer","short":"Majority answer","baseId":"majority-baseline","members":[],"lab":"Other","method":"baseline","variant":null,"harness":null,"ds":{"globalopinionqa":{"sps":0.68956,"idx":-37.8,"lo":-53.3,"hi":-23.8,"pDist":0.53424,"pRank":0.554706,"pRefuse":0.979733,"n":100,"runs":1,"samplesPerQuestion":30,"cost100":0,"latencyP50":null,"badge":"verified","holdoutGap":0.018063,"holdoutLimit":0.1066,"topicIdx":{"politics":-55,"economy":-29.9,"technology":-17,"science-health":-5.9,"identity":-14.9,"society":56.7,"global":-40.2},"topicN":{"politics":31,"economy":6,"technology":6,"science-health":5,"identity":9,"society":1,"global":42},"configId":"baseline--majority-baseline--tdefault--tplcurrent--e36ac49c"},"gss":{"sps":0.721147,"idx":-5.4,"lo":-18.1,"hi":6.6,"pDist":0.541722,"pRank":0.649668,"pRefuse":0.972051,"n":75,"runs":1,"samplesPerQuestion":30,"cost100":0,"latencyP50":0,"badge":"verified","holdoutGap":0.049289,"holdoutLimit":0.129099,"topicIdx":{"politics":9.6,"economy":-28.7,"technology":-70.1,"science-health":27.9,"identity":-12.1,"society":-3.4,"global":-47.4},"topicN":{"politics":24,"economy":5,"technology":5,"science-health":8,"identity":19,"society":12,"global":2},"configId":"baseline--majority-baseline--tdefault--tplcurrent--9194f366"},"ntia":{"sps":0.664986,"idx":-58.1,"lo":-85.1,"hi":-30.1,"pDist":0.564577,"pRank":0.43038,"pRefuse":1,"n":79,"runs":1,"samplesPerQuestion":30,"cost100":0,"latencyP50":0,"badge":"flagged","holdoutGap":0.104719,"holdoutLimit":0.089803,"topicIdx":{"economy":-149.7,"technology":-54.5},"topicN":{"economy":4,"technology":66},"configId":"baseline--majority-baseline--tdefault--tplcurrent--c0bf4946"},"opinionsqa":{"sps":0.704504,"idx":-24.7,"lo":-29.1,"hi":-20.5,"pDist":0.508148,"pRank":0.615837,"pRefuse":0.989527,"n":684,"runs":15,"samplesPerQuestion":30,"cost100":0,"latencyP50":null,"badge":"verified","holdoutGap":0.001061,"holdoutLimit":0.05,"topicIdx":{"politics":-6.4,"economy":-28.6,"technology":-25.1,"science-health":-20,"identity":-26.7,"society":-38.7,"global":-32.6},"topicN":{"politics":80,"economy":73,"technology":146,"science-health":140,"identity":116,"society":82,"global":47},"configId":"baseline--majority-baseline--tdefault--tplcurrent--04e61fe9"},"subpop":{"sps":0.672732,"idx":-35.6,"lo":-44,"hi":-27.8,"pDist":0.466513,"pRank":0.576152,"pRefuse":0.975532,"n":200,"runs":1,"samplesPerQuestion":30,"cost100":0,"latencyP50":null,"badge":"verified","holdoutGap":0.00298,"holdoutLimit":0.068041,"topicIdx":{"politics":-27.2,"economy":-52.5,"technology":-39.3,"science-health":-38.5,"identity":-40.6,"society":-23.2,"global":-31.1},"topicN":{"politics":42,"economy":5,"technology":65,"science-health":4,"identity":56,"society":18,"global":10},"configId":"baseline--majority-baseline--tdefault--tplcurrent--bd8273b6"}},"coverage":4,"index":-25.9,"indexLo":-31.2,"indexHi":-20.6,"cost100":0,"latencyP50":0,"flagged":1,"runs":19}},"findings":{"sps_convention":"sps","generated_from":"leaderboard-results/ (recomputed per-question rows)","temperature_sweep":[{"model":"Claude Haiku 4.5","temperature":0.3,"sps":0.814418,"n_runs":1,"dataset":"opinionsqa"},{"model":"Claude Haiku 4.5","temperature":0.5,"sps":0.821033,"n_runs":1,"dataset":"opinionsqa"},{"model":"Claude Haiku 4.5","temperature":0.7,"sps":0.82496,"n_runs":1,"dataset":"opinionsqa"},{"model":"Claude Haiku 4.5","temperature":0.85,"sps":0.826212,"n_runs":1,"dataset":"opinionsqa"},{"model":"Claude Haiku 4.5","temperature":1,"sps":0.830237,"n_runs":1,"dataset":"opinionsqa"},{"model":"GPT-4o-mini","temperature":0.3,"sps":0.791967,"n_runs":1,"dataset":"opinionsqa"},{"model":"GPT-4o-mini","temperature":0.5,"sps":0.798345,"n_runs":1,"dataset":"opinionsqa"},{"model":"GPT-4o-mini","temperature":0.7,"sps":0.801773,"n_runs":1,"dataset":"opinionsqa"},{"model":"GPT-4o-mini","temperature":0.85,"sps":0.803073,"n_runs":1,"dataset":"opinionsqa"},{"model":"GPT-4o-mini","temperature":1,"sps":0.807262,"n_runs":1,"dataset":"opinionsqa"},{"model":"Gemini Flash Lite","temperature":0.3,"sps":0.7957,"n_runs":1,"dataset":"opinionsqa"},{"model":"Gemini Flash Lite","temperature":0.5,"sps":0.805996,"n_runs":1,"dataset":"opinionsqa"},{"model":"Gemini Flash Lite","temperature":0.7,"sps":0.815377,"n_runs":1,"dataset":"opinionsqa"},{"model":"Gemini Flash Lite","temperature":0.85,"sps":0.824115,"n_runs":1,"dataset":"opinionsqa"},{"model":"Gemini Flash Lite","temperature":1,"sps":0.829589,"n_runs":1,"dataset":"opinionsqa"},{"model":"Gemini Flash Lite","temperature":1.2,"sps":0.835401,"n_runs":1,"dataset":"opinionsqa"},{"model":"Gemini Flash Lite","temperature":1.5,"sps":0.834074,"n_runs":1,"dataset":"opinionsqa"},{"model":"Gemini Flash Lite","temperature":1.8,"sps":0.840725,"n_runs":1,"dataset":"opinionsqa"},{"model":"Gemini Flash Lite","temperature":2,"sps":0.840649,"n_runs":1,"dataset":"opinionsqa"}],"ensemble_comparison":[{"dataset":"globalopinionqa","ensemble_model":"Ensemble: Haiku 4.5, Gemini 2.5 Flash-Lite, GPT-4o mini (Althing)","best_single_model":"Althing (Qwen 3.8 Flash)","best_single_framework":"product","best_single_sps":0.799456,"ensemble_sps":0.810184,"improvement":0.010728,"n_questions":100,"best_member_model":"Althing (GPT-4o-mini)","best_member_sps":0.778334,"improvement_over_members":0.03185,"random_baseline_sps":0.774017,"random_baseline_n":100},{"dataset":"globalopinionqa","ensemble_model":"Ensemble: Qwen 3.8 Flash (Althing), Sonnet 4 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","best_single_model":"Althing (Qwen 3.8 Flash)","best_single_framework":"product","best_single_sps":0.799456,"ensemble_sps":0.830727,"improvement":0.031271,"n_questions":100,"best_member_model":"Althing (Qwen 3.8 Flash)","best_member_sps":0.799456,"improvement_over_members":0.031271,"random_baseline_sps":0.774017,"random_baseline_n":100},{"dataset":"globalopinionqa","ensemble_model":"Ensemble: Haiku 5.5 (Althing), Llama 3.3 70B, Gemini 2.5 Flash-Lite","best_single_model":"Althing (Qwen 3.8 Flash)","best_single_framework":"product","best_single_sps":0.799456,"ensemble_sps":0.818198,"improvement":0.018742,"n_questions":100,"best_member_model":"Althing (Haiku 5.5)","best_member_sps":0.774391,"improvement_over_members":0.043807,"random_baseline_sps":0.774017,"random_baseline_n":100},{"dataset":"globalopinionqa","ensemble_model":"Ensemble: Haiku 5.5 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","best_single_model":"Althing (Qwen 3.8 Flash)","best_single_framework":"product","best_single_sps":0.799456,"ensemble_sps":0.825182,"improvement":0.025726,"n_questions":100,"best_member_model":"DeepSeek V4.1 Flash","best_member_sps":0.79398,"improvement_over_members":0.031202,"random_baseline_sps":0.774017,"random_baseline_n":100},{"dataset":"globalopinionqa","ensemble_model":"Ensemble: Qwen 3.8 Flash, DeepSeek V4.1 Flash, Gemini 3.5 Flash-Lite (raw)","best_single_model":"Althing (Qwen 3.8 Flash)","best_single_framework":"product","best_single_sps":0.799456,"ensemble_sps":0.817169,"improvement":0.017713,"n_questions":100,"best_member_model":"DeepSeek V4.1 Flash","best_member_sps":0.79398,"improvement_over_members":0.023189,"random_baseline_sps":0.774017,"random_baseline_n":100},{"dataset":"gss","ensemble_model":"Ensemble: Haiku 4.5, Gemini 2.5 Flash-Lite, GPT-4o mini (Althing)","best_single_model":"Althing (Qwen 3.8 Flash)","best_single_framework":"product","best_single_sps":0.849052,"ensemble_sps":0.816146,"improvement":-0.032906,"n_questions":75,"best_member_model":"Althing (Gemini Flash Lite)","best_member_sps":0.818377,"improvement_over_members":-0.002231,"random_baseline_sps":0.73538,"random_baseline_n":75},{"dataset":"gss","ensemble_model":"Ensemble: Qwen 3.8 Flash (Althing), Sonnet 4 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","best_single_model":"Althing (Qwen 3.8 Flash)","best_single_framework":"product","best_single_sps":0.849052,"ensemble_sps":0.863134,"improvement":0.014082,"n_questions":75,"best_member_model":"Althing (Qwen 3.8 Flash)","best_member_sps":0.849052,"improvement_over_members":0.014082,"random_baseline_sps":0.73538,"random_baseline_n":75},{"dataset":"gss","ensemble_model":"Ensemble: Haiku 5.5 (Althing), Llama 3.3 70B, Gemini 2.5 Flash-Lite","best_single_model":"Althing (Qwen 3.8 Flash)","best_single_framework":"product","best_single_sps":0.849052,"ensemble_sps":0.847516,"improvement":-0.001536,"n_questions":75,"best_member_model":"Llama 3.3 70B","best_member_sps":0.814656,"improvement_over_members":0.03286,"random_baseline_sps":0.73538,"random_baseline_n":75},{"dataset":"gss","ensemble_model":"Ensemble: Haiku 5.5 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","best_single_model":"Althing (Qwen 3.8 Flash)","best_single_framework":"product","best_single_sps":0.849052,"ensemble_sps":0.851459,"improvement":0.002407,"n_questions":75,"best_member_model":"DeepSeek V4.1 Flash","best_member_sps":0.833673,"improvement_over_members":0.017786,"random_baseline_sps":0.73538,"random_baseline_n":75},{"dataset":"gss","ensemble_model":"Ensemble: Qwen 3.8 Flash, DeepSeek V4.1 Flash, Gemini 3.5 Flash-Lite (raw)","best_single_model":"Althing (Qwen 3.8 Flash)","best_single_framework":"product","best_single_sps":0.849052,"ensemble_sps":0.844613,"improvement":-0.004439,"n_questions":75,"best_member_model":"Qwen 3.8 Flash","best_member_sps":0.83603,"improvement_over_members":0.008583,"random_baseline_sps":0.73538,"random_baseline_n":75},{"dataset":"opinionsqa","ensemble_model":"Ensemble: Haiku 4.5, Gemini 2.5 Flash-Lite, GPT-4o mini (Althing)","best_single_model":"DeepSeek V4.1 Flash","best_single_framework":"raw","best_single_sps":0.873957,"ensemble_sps":0.87124,"improvement":-0.002717,"n_questions":100,"best_member_model":"Althing (Haiku 4.5)","best_member_sps":0.814418,"improvement_over_members":0.056822,"random_baseline_sps":0.762852,"random_baseline_n":684},{"dataset":"opinionsqa","ensemble_model":"Ensemble: Qwen 3.8 Flash (Althing), Sonnet 4 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","best_single_model":"DeepSeek V4.1 Flash","best_single_framework":"raw","best_single_sps":0.873957,"ensemble_sps":0.912853,"improvement":0.038896,"n_questions":100,"best_member_model":"DeepSeek V4.1 Flash","best_member_sps":0.873957,"improvement_over_members":0.038896,"random_baseline_sps":0.762852,"random_baseline_n":684},{"dataset":"opinionsqa","ensemble_model":"Ensemble: Haiku 5.5 (Althing), Llama 3.3 70B, Gemini 2.5 Flash-Lite","best_single_model":"DeepSeek V4.1 Flash","best_single_framework":"raw","best_single_sps":0.873957,"ensemble_sps":0.897395,"improvement":0.023438,"n_questions":100,"best_member_model":"Gemini Flash Lite","best_member_sps":0.825897,"improvement_over_members":0.071498,"random_baseline_sps":0.762852,"random_baseline_n":684},{"dataset":"opinionsqa","ensemble_model":"Ensemble: Haiku 5.5 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","best_single_model":"DeepSeek V4.1 Flash","best_single_framework":"raw","best_single_sps":0.873957,"ensemble_sps":0.905567,"improvement":0.03161,"n_questions":100,"best_member_model":"DeepSeek V4.1 Flash","best_member_sps":0.873957,"improvement_over_members":0.03161,"random_baseline_sps":0.762852,"random_baseline_n":684},{"dataset":"opinionsqa","ensemble_model":"Ensemble: Qwen 3.8 Flash, DeepSeek V4.1 Flash, Gemini 3.5 Flash-Lite (raw)","best_single_model":"DeepSeek V4.1 Flash","best_single_framework":"raw","best_single_sps":0.873957,"ensemble_sps":0.890868,"improvement":0.016911,"n_questions":100,"best_member_model":"DeepSeek V4.1 Flash","best_member_sps":0.873957,"improvement_over_members":0.016911,"random_baseline_sps":0.762852,"random_baseline_n":684},{"dataset":"subpop","ensemble_model":"Ensemble: Haiku 4.5, Gemini 2.5 Flash-Lite, GPT-4o mini (Althing)","best_single_model":"DeepSeek V4.1 Flash","best_single_framework":"raw","best_single_sps":0.83753,"ensemble_sps":0.823337,"improvement":-0.014193,"n_questions":100,"best_member_model":"Althing (Haiku 4.5)","best_member_sps":0.770122,"improvement_over_members":0.053215,"random_baseline_sps":0.757397,"random_baseline_n":200},{"dataset":"subpop","ensemble_model":"Ensemble: Qwen 3.8 Flash (Althing), Sonnet 4 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","best_single_model":"DeepSeek V4.1 Flash","best_single_framework":"raw","best_single_sps":0.83753,"ensemble_sps":0.864985,"improvement":0.027455,"n_questions":100,"best_member_model":"DeepSeek V4.1 Flash","best_member_sps":0.83753,"improvement_over_members":0.027455,"random_baseline_sps":0.757397,"random_baseline_n":200},{"dataset":"subpop","ensemble_model":"Ensemble: Haiku 5.5 (Althing), Llama 3.3 70B, Gemini 2.5 Flash-Lite","best_single_model":"DeepSeek V4.1 Flash","best_single_framework":"raw","best_single_sps":0.83753,"ensemble_sps":0.845827,"improvement":0.008297,"n_questions":100,"best_member_model":"Althing (Haiku 5.5)","best_member_sps":0.783362,"improvement_over_members":0.062465,"random_baseline_sps":0.757397,"random_baseline_n":200},{"dataset":"subpop","ensemble_model":"Ensemble: Haiku 5.5 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","best_single_model":"DeepSeek V4.1 Flash","best_single_framework":"raw","best_single_sps":0.83753,"ensemble_sps":0.856103,"improvement":0.018573,"n_questions":100,"best_member_model":"DeepSeek V4.1 Flash","best_member_sps":0.83753,"improvement_over_members":0.018573,"random_baseline_sps":0.757397,"random_baseline_n":200},{"dataset":"subpop","ensemble_model":"Ensemble: Qwen 3.8 Flash, DeepSeek V4.1 Flash, Gemini 3.5 Flash-Lite (raw)","best_single_model":"DeepSeek V4.1 Flash","best_single_framework":"raw","best_single_sps":0.83753,"ensemble_sps":0.859209,"improvement":0.021679,"n_questions":100,"best_member_model":"DeepSeek V4.1 Flash","best_member_sps":0.83753,"improvement_over_members":0.021679,"random_baseline_sps":0.757397,"random_baseline_n":200}],"ensemble_error_correlation":[{"dataset":"globalopinionqa","ensemble_model":"Ensemble: Haiku 4.5, Gemini 2.5 Flash-Lite, GPT-4o mini (Althing)","n_questions":100,"constituents":["Althing (Haiku 4.5)","Althing (Gemini Flash Lite)","Althing (GPT-4o-mini)"],"pairs":[{"a":"Althing (Haiku 4.5)","b":"Althing (Gemini Flash Lite)","pearson_r":0.509877,"n":100},{"a":"Althing (Haiku 4.5)","b":"Althing (GPT-4o-mini)","pearson_r":0.575954,"n":100},{"a":"Althing (Gemini Flash Lite)","b":"Althing (GPT-4o-mini)","pearson_r":0.497951,"n":100}],"mean_pearson_r":0.527927},{"dataset":"globalopinionqa","ensemble_model":"Ensemble: Qwen 3.8 Flash (Althing), Sonnet 4 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","n_questions":100,"constituents":["Althing (Qwen 3.8 Flash)","Althing (Sonnet 4)","DeepSeek V4.1 Flash","Gemini Flash Lite"],"pairs":[{"a":"Althing (Qwen 3.8 Flash)","b":"Althing (Sonnet 4)","pearson_r":0.622334,"n":100},{"a":"Althing (Qwen 3.8 Flash)","b":"DeepSeek V4.1 Flash","pearson_r":0.449595,"n":100},{"a":"Althing (Qwen 3.8 Flash)","b":"Gemini Flash Lite","pearson_r":0.469593,"n":100},{"a":"Althing (Sonnet 4)","b":"DeepSeek V4.1 Flash","pearson_r":0.482534,"n":100},{"a":"Althing (Sonnet 4)","b":"Gemini Flash Lite","pearson_r":0.608687,"n":100},{"a":"DeepSeek V4.1 Flash","b":"Gemini Flash Lite","pearson_r":0.59256,"n":100}],"mean_pearson_r":0.537551},{"dataset":"globalopinionqa","ensemble_model":"Ensemble: Haiku 5.5 (Althing), Llama 3.3 70B, Gemini 2.5 Flash-Lite","n_questions":100,"constituents":["Althing (Haiku 5.5)","Llama 3.3 70B","Gemini Flash Lite"],"pairs":[{"a":"Althing (Haiku 5.5)","b":"Llama 3.3 70B","pearson_r":0.647427,"n":100},{"a":"Althing (Haiku 5.5)","b":"Gemini Flash Lite","pearson_r":0.455416,"n":100},{"a":"Llama 3.3 70B","b":"Gemini Flash Lite","pearson_r":0.547793,"n":100}],"mean_pearson_r":0.550212},{"dataset":"globalopinionqa","ensemble_model":"Ensemble: Haiku 5.5 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","n_questions":100,"constituents":["Althing (Haiku 5.5)","DeepSeek V4.1 Flash","Gemini Flash Lite"],"pairs":[{"a":"Althing (Haiku 5.5)","b":"DeepSeek V4.1 Flash","pearson_r":0.431316,"n":100},{"a":"Althing (Haiku 5.5)","b":"Gemini Flash Lite","pearson_r":0.455416,"n":100},{"a":"DeepSeek V4.1 Flash","b":"Gemini Flash Lite","pearson_r":0.59256,"n":100}],"mean_pearson_r":0.493097},{"dataset":"globalopinionqa","ensemble_model":"Ensemble: Qwen 3.8 Flash, DeepSeek V4.1 Flash, Gemini 3.5 Flash-Lite (raw)","n_questions":100,"constituents":["Qwen 3.8 Flash","DeepSeek V4.1 Flash","Gemini 3.5 Flash-Lite"],"pairs":[{"a":"Qwen 3.8 Flash","b":"DeepSeek V4.1 Flash","pearson_r":0.496554,"n":100},{"a":"Qwen 3.8 Flash","b":"Gemini 3.5 Flash-Lite","pearson_r":0.461388,"n":100},{"a":"DeepSeek V4.1 Flash","b":"Gemini 3.5 Flash-Lite","pearson_r":0.645707,"n":100}],"mean_pearson_r":0.53455},{"dataset":"gss","ensemble_model":"Ensemble: Haiku 4.5, Gemini 2.5 Flash-Lite, GPT-4o mini (Althing)","n_questions":75,"constituents":["Althing (Haiku 4.5)","Althing (Gemini Flash Lite)","Althing (GPT-4o-mini)"],"pairs":[{"a":"Althing (Haiku 4.5)","b":"Althing (Gemini Flash Lite)","pearson_r":0.601136,"n":75},{"a":"Althing (Haiku 4.5)","b":"Althing (GPT-4o-mini)","pearson_r":0.617184,"n":75},{"a":"Althing (Gemini Flash Lite)","b":"Althing (GPT-4o-mini)","pearson_r":0.680464,"n":75}],"mean_pearson_r":0.632928},{"dataset":"gss","ensemble_model":"Ensemble: Qwen 3.8 Flash (Althing), Sonnet 4 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","n_questions":75,"constituents":["Althing (Qwen 3.8 Flash)","Althing (Sonnet 4)","DeepSeek V4.1 Flash","Gemini Flash Lite"],"pairs":[{"a":"Althing (Qwen 3.8 Flash)","b":"Althing (Sonnet 4)","pearson_r":0.632127,"n":75},{"a":"Althing (Qwen 3.8 Flash)","b":"DeepSeek V4.1 Flash","pearson_r":0.581519,"n":75},{"a":"Althing (Qwen 3.8 Flash)","b":"Gemini Flash Lite","pearson_r":0.461916,"n":75},{"a":"Althing (Sonnet 4)","b":"DeepSeek V4.1 Flash","pearson_r":0.618719,"n":75},{"a":"Althing (Sonnet 4)","b":"Gemini Flash Lite","pearson_r":0.486443,"n":75},{"a":"DeepSeek V4.1 Flash","b":"Gemini Flash Lite","pearson_r":0.398118,"n":75}],"mean_pearson_r":0.529807},{"dataset":"gss","ensemble_model":"Ensemble: Haiku 5.5 (Althing), Llama 3.3 70B, Gemini 2.5 Flash-Lite","n_questions":75,"constituents":["Althing (Haiku 5.5)","Llama 3.3 70B","Gemini Flash Lite"],"pairs":[{"a":"Althing (Haiku 5.5)","b":"Llama 3.3 70B","pearson_r":0.590692,"n":75},{"a":"Althing (Haiku 5.5)","b":"Gemini Flash Lite","pearson_r":0.516712,"n":75},{"a":"Llama 3.3 70B","b":"Gemini Flash Lite","pearson_r":0.531701,"n":75}],"mean_pearson_r":0.546368},{"dataset":"gss","ensemble_model":"Ensemble: Haiku 5.5 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","n_questions":75,"constituents":["Althing (Haiku 5.5)","DeepSeek V4.1 Flash","Gemini Flash Lite"],"pairs":[{"a":"Althing (Haiku 5.5)","b":"DeepSeek V4.1 Flash","pearson_r":0.467622,"n":75},{"a":"Althing (Haiku 5.5)","b":"Gemini Flash Lite","pearson_r":0.516712,"n":75},{"a":"DeepSeek V4.1 Flash","b":"Gemini Flash Lite","pearson_r":0.398118,"n":75}],"mean_pearson_r":0.460817},{"dataset":"gss","ensemble_model":"Ensemble: Qwen 3.8 Flash, DeepSeek V4.1 Flash, Gemini 3.5 Flash-Lite (raw)","n_questions":75,"constituents":["Qwen 3.8 Flash","DeepSeek V4.1 Flash","Gemini 3.5 Flash-Lite"],"pairs":[{"a":"Qwen 3.8 Flash","b":"DeepSeek V4.1 Flash","pearson_r":0.66734,"n":75},{"a":"Qwen 3.8 Flash","b":"Gemini 3.5 Flash-Lite","pearson_r":0.643644,"n":75},{"a":"DeepSeek V4.1 Flash","b":"Gemini 3.5 Flash-Lite","pearson_r":0.499691,"n":75}],"mean_pearson_r":0.603558},{"dataset":"opinionsqa","ensemble_model":"Ensemble: Haiku 4.5, Gemini 2.5 Flash-Lite, GPT-4o mini (Althing)","n_questions":100,"constituents":["Althing (Haiku 4.5)","Althing (Gemini Flash Lite)","Althing (GPT-4o-mini)"],"pairs":[{"a":"Althing (Haiku 4.5)","b":"Althing (Gemini Flash Lite)","pearson_r":0.518,"n":100},{"a":"Althing (Haiku 4.5)","b":"Althing (GPT-4o-mini)","pearson_r":0.359555,"n":100},{"a":"Althing (Gemini Flash Lite)","b":"Althing (GPT-4o-mini)","pearson_r":0.617913,"n":100}],"mean_pearson_r":0.498489},{"dataset":"opinionsqa","ensemble_model":"Ensemble: Qwen 3.8 Flash (Althing), Sonnet 4 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","n_questions":100,"constituents":["Althing (Qwen 3.8 Flash)","Althing (Sonnet 4)","DeepSeek V4.1 Flash","Gemini Flash Lite"],"pairs":[{"a":"Althing (Qwen 3.8 Flash)","b":"Althing (Sonnet 4)","pearson_r":0.562157,"n":100},{"a":"Althing (Qwen 3.8 Flash)","b":"DeepSeek V4.1 Flash","pearson_r":0.398402,"n":100},{"a":"Althing (Qwen 3.8 Flash)","b":"Gemini Flash Lite","pearson_r":0.345451,"n":100},{"a":"Althing (Sonnet 4)","b":"DeepSeek V4.1 Flash","pearson_r":0.467562,"n":100},{"a":"Althing (Sonnet 4)","b":"Gemini Flash Lite","pearson_r":0.389553,"n":100},{"a":"DeepSeek V4.1 Flash","b":"Gemini Flash Lite","pearson_r":0.300087,"n":100}],"mean_pearson_r":0.410535},{"dataset":"opinionsqa","ensemble_model":"Ensemble: Haiku 5.5 (Althing), Llama 3.3 70B, Gemini 2.5 Flash-Lite","n_questions":100,"constituents":["Althing (Haiku 5.5)","Llama 3.3 70B","Gemini Flash Lite"],"pairs":[{"a":"Althing (Haiku 5.5)","b":"Llama 3.3 70B","pearson_r":0.432466,"n":100},{"a":"Althing (Haiku 5.5)","b":"Gemini Flash Lite","pearson_r":0.232156,"n":100},{"a":"Llama 3.3 70B","b":"Gemini Flash Lite","pearson_r":0.485982,"n":100}],"mean_pearson_r":0.383535},{"dataset":"opinionsqa","ensemble_model":"Ensemble: Haiku 5.5 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","n_questions":100,"constituents":["Althing (Haiku 5.5)","DeepSeek V4.1 Flash","Gemini Flash Lite"],"pairs":[{"a":"Althing (Haiku 5.5)","b":"DeepSeek V4.1 Flash","pearson_r":0.421599,"n":100},{"a":"Althing (Haiku 5.5)","b":"Gemini Flash Lite","pearson_r":0.232156,"n":100},{"a":"DeepSeek V4.1 Flash","b":"Gemini Flash Lite","pearson_r":0.300087,"n":100}],"mean_pearson_r":0.317947},{"dataset":"opinionsqa","ensemble_model":"Ensemble: Qwen 3.8 Flash, DeepSeek V4.1 Flash, Gemini 3.5 Flash-Lite (raw)","n_questions":100,"constituents":["Qwen 3.8 Flash","DeepSeek V4.1 Flash","Gemini 3.5 Flash-Lite"],"pairs":[{"a":"Qwen 3.8 Flash","b":"DeepSeek V4.1 Flash","pearson_r":0.321593,"n":100},{"a":"Qwen 3.8 Flash","b":"Gemini 3.5 Flash-Lite","pearson_r":0.329067,"n":100},{"a":"DeepSeek V4.1 Flash","b":"Gemini 3.5 Flash-Lite","pearson_r":0.55817,"n":100}],"mean_pearson_r":0.402943},{"dataset":"subpop","ensemble_model":"Ensemble: Haiku 4.5, Gemini 2.5 Flash-Lite, GPT-4o mini (Althing)","n_questions":100,"constituents":["Althing (Haiku 4.5)","Althing (Gemini Flash Lite)","Althing (GPT-4o-mini)"],"pairs":[{"a":"Althing (Haiku 4.5)","b":"Althing (Gemini Flash Lite)","pearson_r":0.230635,"n":100},{"a":"Althing (Haiku 4.5)","b":"Althing (GPT-4o-mini)","pearson_r":0.388237,"n":100},{"a":"Althing (Gemini Flash Lite)","b":"Althing (GPT-4o-mini)","pearson_r":0.439084,"n":100}],"mean_pearson_r":0.352652},{"dataset":"subpop","ensemble_model":"Ensemble: Qwen 3.8 Flash (Althing), Sonnet 4 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","n_questions":100,"constituents":["Althing (Qwen 3.8 Flash)","Althing (Sonnet 4)","DeepSeek V4.1 Flash","Gemini Flash Lite"],"pairs":[{"a":"Althing (Qwen 3.8 Flash)","b":"Althing (Sonnet 4)","pearson_r":0.684642,"n":100},{"a":"Althing (Qwen 3.8 Flash)","b":"DeepSeek V4.1 Flash","pearson_r":0.586963,"n":100},{"a":"Althing (Qwen 3.8 Flash)","b":"Gemini Flash Lite","pearson_r":0.562438,"n":100},{"a":"Althing (Sonnet 4)","b":"DeepSeek V4.1 Flash","pearson_r":0.468195,"n":100},{"a":"Althing (Sonnet 4)","b":"Gemini Flash Lite","pearson_r":0.49032,"n":100},{"a":"DeepSeek V4.1 Flash","b":"Gemini Flash Lite","pearson_r":0.343439,"n":100}],"mean_pearson_r":0.522666},{"dataset":"subpop","ensemble_model":"Ensemble: Haiku 5.5 (Althing), Llama 3.3 70B, Gemini 2.5 Flash-Lite","n_questions":100,"constituents":["Althing (Haiku 5.5)","Llama 3.3 70B","Gemini Flash Lite"],"pairs":[{"a":"Althing (Haiku 5.5)","b":"Llama 3.3 70B","pearson_r":0.288709,"n":100},{"a":"Althing (Haiku 5.5)","b":"Gemini Flash Lite","pearson_r":0.229848,"n":100},{"a":"Llama 3.3 70B","b":"Gemini Flash Lite","pearson_r":0.411741,"n":100}],"mean_pearson_r":0.310099},{"dataset":"subpop","ensemble_model":"Ensemble: Haiku 5.5 (Althing), DeepSeek V4.1 Flash, Gemini 2.5 Flash-Lite","n_questions":100,"constituents":["Althing (Haiku 5.5)","DeepSeek V4.1 Flash","Gemini Flash Lite"],"pairs":[{"a":"Althing (Haiku 5.5)","b":"DeepSeek V4.1 Flash","pearson_r":0.178344,"n":100},{"a":"Althing (Haiku 5.5)","b":"Gemini Flash Lite","pearson_r":0.229848,"n":100},{"a":"DeepSeek V4.1 Flash","b":"Gemini Flash Lite","pearson_r":0.343439,"n":100}],"mean_pearson_r":0.250544},{"dataset":"subpop","ensemble_model":"Ensemble: Qwen 3.8 Flash, DeepSeek V4.1 Flash, Gemini 3.5 Flash-Lite (raw)","n_questions":100,"constituents":["Qwen 3.8 Flash","DeepSeek V4.1 Flash","Gemini 3.5 Flash-Lite"],"pairs":[{"a":"Qwen 3.8 Flash","b":"DeepSeek V4.1 Flash","pearson_r":0.467256,"n":100},{"a":"Qwen 3.8 Flash","b":"Gemini 3.5 Flash-Lite","pearson_r":0.507591,"n":100},{"a":"DeepSeek V4.1 Flash","b":"Gemini 3.5 Flash-Lite","pearson_r":0.380358,"n":100}],"mean_pearson_r":0.451735}],"conditioning_results":[{"attribute":"POLPARTY","group":"Republican","group_raw":"Republican","p_dist":0.683448,"p_cond":0.105296,"n_replications":4,"p_cond_std":0.019912},{"attribute":"POLPARTY","group":"Democrat","group_raw":"Democrat","p_dist":0.667307,"p_cond":0.05181,"n_replications":4,"p_cond_std":0.021901},{"attribute":"INCOME","group":"<$30K","group_raw":"Less than $30,000","p_dist":0.652378,"p_cond":0.071025,"n_replications":4,"p_cond_std":0.014041},{"attribute":"INCOME","group":"$100K+","group_raw":"$100,000 or more","p_dist":0.673218,"p_cond":0.044288,"n_replications":4,"p_cond_std":0.028961},{"attribute":"EDUCATION","group":"Less than HS","group_raw":"Less than high school","p_dist":0.652389,"p_cond":0.086877,"n_replications":4,"p_cond_std":0.028208},{"attribute":"EDUCATION","group":"College graduate","group_raw":"College graduate/some postgrad","p_dist":0.683954,"p_cond":0.040417,"n_replications":4,"p_cond_std":0.022973}],"conditioning_extended":[{"attribute":"CREGION","group":"South","group_raw":"South","p_dist":0.658431,"p_cond":0.059878,"n_replications":4,"p_cond_std":0.022178},{"attribute":"CREGION","group":"Northeast","group_raw":"Northeast","p_dist":0.677018,"p_cond":0.055926,"n_replications":4,"p_cond_std":0.03418},{"attribute":"SEX","group":"Male","group_raw":"Male","p_dist":0.68288,"p_cond":0.051707,"n_replications":4,"p_cond_std":0.038139},{"attribute":"SEX","group":"Female","group_raw":"Female","p_dist":0.656583,"p_cond":0.05143,"n_replications":4,"p_cond_std":0.032953}],"template_comparison":[{"template":"current","sps":0.725414,"n_runs":1},{"template":"demo","sps":0.667463,"n_runs":1},{"template":"minimal","sps":0.648169,"n_runs":1},{"template":"values","sps":0.637425,"n_runs":1}],"elicitation_comparison":[{"model":"Althing (GPT-6 Luna)","framework":"product","dataset":"globalopinionqa","elicitation":"structured","template":"structured","n_questions":100,"samples_per_question":30,"refusal_detector_version":3,"natural":{"sps":0.762266,"p_dist":0.665298,"p_rank":0.641767,"p_refuse":0.979733,"parse_failure_rate":0.008},"structured":{"sps":0.76475,"p_dist":0.653872,"p_rank":0.660646,"p_refuse":0.979733,"parse_failure_rate":0},"delta":{"sps":0.002484,"p_dist":-0.011426,"p_rank":0.018879,"p_refuse":0,"parse_failure_rate":-0.008}},{"model":"Althing (Haiku 5.5)","framework":"product","dataset":"globalopinionqa","elicitation":"structured","template":"structured","n_questions":100,"samples_per_question":30,"refusal_detector_version":4,"natural":{"sps":0.774391,"p_dist":0.660135,"p_rank":0.684175,"p_refuse":0.978863,"parse_failure_rate":0.006667},"structured":{"sps":0.75311,"p_dist":0.609389,"p_rank":0.670208,"p_refuse":0.979733,"parse_failure_rate":0.000333},"delta":{"sps":-0.021281,"p_dist":-0.050746,"p_rank":-0.013967,"p_refuse":0.00087,"parse_failure_rate":-0.006334}},{"model":"Althing (GPT-6 Luna)","framework":"product","dataset":"opinionsqa","elicitation":"structured","template":"structured","n_questions":100,"samples_per_question":30,"refusal_detector_version":3,"natural":{"sps":0.832773,"p_dist":0.745826,"p_rank":0.758221,"p_refuse":0.994272,"parse_failure_rate":0.005},"structured":{"sps":0.815631,"p_dist":0.728301,"p_rank":0.72432,"p_refuse":0.994272,"parse_failure_rate":0},"delta":{"sps":-0.017142,"p_dist":-0.017525,"p_rank":-0.033901,"p_refuse":0,"parse_failure_rate":-0.005}},{"model":"Althing (Haiku 5.5)","framework":"product","dataset":"opinionsqa","elicitation":"structured","template":"structured","n_questions":100,"samples_per_question":30,"refusal_detector_version":4,"natural":{"sps":0.822803,"p_dist":0.700423,"p_rank":0.775571,"p_refuse":0.992416,"parse_failure_rate":0.000667},"structured":{"sps":0.834444,"p_dist":0.718182,"p_rank":0.790879,"p_refuse":0.994272,"parse_failure_rate":0},"delta":{"sps":0.011641,"p_dist":0.017759,"p_rank":0.015308,"p_refuse":0.001856,"parse_failure_rate":-0.000667}},{"model":"Althing (GPT-6 Luna)","framework":"product","dataset":"subpop","elicitation":"structured","template":"structured","n_questions":100,"samples_per_question":30,"refusal_detector_version":3,"natural":{"sps":0.802001,"p_dist":0.696016,"p_rank":0.730057,"p_refuse":0.979931,"parse_failure_rate":0.003333},"structured":{"sps":0.790997,"p_dist":0.666088,"p_rank":0.726974,"p_refuse":0.979931,"parse_failure_rate":0},"delta":{"sps":-0.011004,"p_dist":-0.029928,"p_rank":-0.003083,"p_refuse":0,"parse_failure_rate":-0.003333}},{"model":"Althing (Haiku 5.5)","framework":"product","dataset":"subpop","elicitation":"structured","template":"structured","n_questions":100,"samples_per_question":30,"refusal_detector_version":4,"natural":{"sps":0.783362,"p_dist":0.645295,"p_rank":0.730853,"p_refuse":0.973938,"parse_failure_rate":0.008333},"structured":{"sps":0.790644,"p_dist":0.635892,"p_rank":0.756109,"p_refuse":0.979931,"parse_failure_rate":0},"delta":{"sps":0.007282,"p_dist":-0.009403,"p_rank":0.025256,"p_refuse":0.005993,"parse_failure_rate":-0.008333}}],"nonresponse_fidelity":[{"provider":"Muse Glimmer 30B","framework":"raw","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.697013,"mean_model_nonresponse_mass":0.743112,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.738128,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_CONFINAN","model_mass":1,"human_mass":0.01314,"gap":0.98686},{"key":"GSS_FINRELA","model_mass":1,"human_mass":0.014308,"gap":0.985692},{"key":"GSS_XMARSEX","model_mass":1,"human_mass":0.01613,"gap":0.98387},{"key":"GSS_PORNLAW","model_mass":1,"human_mass":0.017909,"gap":0.982091},{"key":"GSS_CLASS","model_mass":1,"human_mass":0.02169,"gap":0.97831}],"top_option_overselected":[{"key":"GSS_CONFINAN","text":"I am going to name an institution in this country. As far as the people running this institution are concerned, would you say you have a great deal of confidence, only some confidence, or hardly any confidence at all in banks and financial institutions?","model_mass":1,"human_mass":0.006557,"gap":0.993443},{"key":"GSS_XMARSEX","text":"What is your opinion about a married person having sexual relations with someone other than the marriage partner--is it always wrong, almost always wrong, wrong only sometimes, or not wrong at all?","model_mass":1,"human_mass":0.008034,"gap":0.991966},{"key":"GSS_PORNLAW","text":"Which of these statements comes closest to your feelings about pornography laws? There should be laws against the distribution of pornography whatever the age; there should be laws against the distribution of pornography to persons under 18; or there should be no laws forbidding the distribution of pornography.","model_mass":1,"human_mass":0.008919,"gap":0.991081},{"key":"GSS_CLASS","text":"If you were asked to use one of four names for your social class, which would you say you belong in: the lower class, the working class, the middle class, or the upper class?","model_mass":1,"human_mass":0.010781,"gap":0.989219},{"key":"GSS_CONJUDGE","text":"I am going to name an institution in this country. As far as the people running this institution are concerned, would you say you have a great deal of confidence, only some confidence, or hardly any confidence at all in the U.S. Supreme Court?","model_mass":1,"human_mass":0.01429,"gap":0.98571}]},{"provider":"Gemini 3.8 Flash","framework":"raw","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.678629,"mean_model_nonresponse_mass":0.718588,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.718588,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_HEALTH","model_mass":1,"human_mass":0.001396,"gap":0.998604},{"key":"GSS_HAPPY","model_mass":1,"human_mass":0.007231,"gap":0.992769},{"key":"GSS_SATFIN","model_mass":1,"human_mass":0.008228,"gap":0.991772},{"key":"GSS_FINRELA","model_mass":1,"human_mass":0.014308,"gap":0.985692},{"key":"GSS_LIFE","model_mass":1,"human_mass":0.014754,"gap":0.985246}],"top_option_overselected":[{"key":"GSS_HEALTH","text":"Would you say your own health, in general, is excellent, good, fair, or poor?","model_mass":1,"human_mass":0.0007,"gap":0.9993},{"key":"GSS_HAPPY","text":"Taken all together, how would you say things are these days--would you say that you are very happy, pretty happy, or not too happy?","model_mass":1,"human_mass":0.003587,"gap":0.996413},{"key":"GSS_SATFIN","text":"We are interested in how people are getting along financially these days. So far as you and your family are concerned, would you say that you are pretty well satisfied with your present financial situation, more or less satisfied, or not satisfied at all?","model_mass":1,"human_mass":0.004083,"gap":0.995917},{"key":"GSS_FINRELA","text":"Compared with American families in general, would you say your family income is far below average, below average, average, above average, or far above average?","model_mass":1,"human_mass":0.007148,"gap":0.992852},{"key":"GSS_LIFE","text":"In general, do you find life exciting, pretty routine, or dull?","model_mass":1,"human_mass":0.007345,"gap":0.992655}]},{"provider":"GPT-6 Luna","framework":"raw","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.308814,"mean_model_nonresponse_mass":0.321775,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.321775,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_HEALTH","model_mass":1,"human_mass":0.001396,"gap":0.998604},{"key":"GSS_FINRELA","model_mass":1,"human_mass":0.014308,"gap":0.985692},{"key":"GSS_CLASS","model_mass":1,"human_mass":0.02169,"gap":0.97831},{"key":"GSS_OWNGUN","model_mass":1,"human_mass":0.029577,"gap":0.970423},{"key":"GSS_TAX","model_mass":1,"human_mass":0.052342,"gap":0.947658}],"top_option_overselected":[{"key":"GSS_HEALTH","text":"Would you say your own health, in general, is excellent, good, fair, or poor?","model_mass":1,"human_mass":0.0007,"gap":0.9993},{"key":"GSS_FINRELA","text":"Compared with American families in general, would you say your family income is far below average, below average, average, above average, or far above average?","model_mass":1,"human_mass":0.007148,"gap":0.992852},{"key":"GSS_CLASS","text":"If you were asked to use one of four names for your social class, which would you say you belong in: the lower class, the working class, the middle class, or the upper class?","model_mass":1,"human_mass":0.010781,"gap":0.989219},{"key":"GSS_OWNGUN","text":"Do you happen to have in your home or garage any guns or revolvers?","model_mass":1,"human_mass":0.014678,"gap":0.985322},{"key":"GSS_TAX","text":"Do you consider the amount of federal income tax which you have to pay as too high, about right, or too low?","model_mass":1,"human_mass":0.025797,"gap":0.974203}]},{"provider":"GPT-6 Luna","framework":"raw","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.249998,"mean_model_nonresponse_mass":0.243111,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.243111,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_HEALTH","model_mass":1,"human_mass":0.001396,"gap":0.998604},{"key":"GSS_SATFIN","model_mass":1,"human_mass":0.008228,"gap":0.991772},{"key":"GSS_FINRELA","model_mass":1,"human_mass":0.014308,"gap":0.985692},{"key":"GSS_CONARMY","model_mass":1,"human_mass":0.015319,"gap":0.984681},{"key":"GSS_CLASS","model_mass":1,"human_mass":0.02169,"gap":0.97831}],"top_option_overselected":[{"key":"GSS_HEALTH","text":"Would you say your own health, in general, is excellent, good, fair, or poor?","model_mass":1,"human_mass":0.0007,"gap":0.9993},{"key":"GSS_SATFIN","text":"We are interested in how people are getting along financially these days. So far as you and your family are concerned, would you say that you are pretty well satisfied with your present financial situation, more or less satisfied, or not satisfied at all?","model_mass":1,"human_mass":0.004083,"gap":0.995917},{"key":"GSS_FINRELA","text":"Compared with American families in general, would you say your family income is far below average, below average, average, above average, or far above average?","model_mass":1,"human_mass":0.007148,"gap":0.992852},{"key":"GSS_CONARMY","text":"I am going to name an institution in this country. As far as the people running this institution are concerned, would you say you have a great deal of confidence, only some confidence, or hardly any confidence at all in the military?","model_mass":1,"human_mass":0.007641,"gap":0.992359},{"key":"GSS_CLASS","text":"If you were asked to use one of four names for your social class, which would you say you belong in: the lower class, the working class, the middle class, or the upper class?","model_mass":1,"human_mass":0.010781,"gap":0.989219}]},{"provider":"Claude Haiku 5.5","framework":"raw","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.230528,"mean_model_nonresponse_mass":0.22653,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.199735,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_HEALTH","model_mass":1,"human_mass":0.001396,"gap":0.998604},{"key":"GSS_FINRELA","model_mass":1,"human_mass":0.014308,"gap":0.985692},{"key":"GSS_SATFIN","model_mass":0.9667,"human_mass":0.008228,"gap":0.958472},{"key":"GSS_CAPPUN","model_mass":1,"human_mass":0.068223,"gap":0.931777},{"key":"GSS_ABANY","model_mass":1,"human_mass":0.068418,"gap":0.931582}],"top_option_overselected":[{"key":"GSS_HEALTH","text":"Would you say your own health, in general, is excellent, good, fair, or poor?","model_mass":1,"human_mass":0.0007,"gap":0.9993},{"key":"GSS_CAPPUN","text":"Do you favor or oppose the death penalty for persons convicted of murder?","model_mass":1,"human_mass":0.033495,"gap":0.966505},{"key":"GSS_ABANY","text":"Please tell me whether or not you think it should be possible for a pregnant woman to obtain a legal abortion if the woman wants it for any reason?","model_mass":1,"human_mass":0.033588,"gap":0.966412},{"key":"GSS_SATFIN","text":"We are interested in how people are getting along financially these days. So far as you and your family are concerned, would you say that you are pretty well satisfied with your present financial situation, more or less satisfied, or not satisfied at all?","model_mass":0.9667,"human_mass":0.004083,"gap":0.962617},{"key":"GSS_ABNOMORE","text":"Please tell me whether or not you think it should be possible for a pregnant woman to obtain a legal abortion if she is married and does not want any more children?","model_mass":1,"human_mass":0.04051,"gap":0.95949}]},{"provider":"Claude Sonnet 4","framework":"raw","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.229692,"mean_model_nonresponse_mass":0.226282,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.136473,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_HEALTH","model_mass":1,"human_mass":0.001396,"gap":0.998604},{"key":"GSS_SATFIN","model_mass":1,"human_mass":0.008228,"gap":0.991772},{"key":"GSS_FINRELA","model_mass":1,"human_mass":0.014308,"gap":0.985692},{"key":"GSS_CLASS","model_mass":1,"human_mass":0.02169,"gap":0.97831},{"key":"GSS_TAX","model_mass":1,"human_mass":0.052342,"gap":0.947658}],"top_option_overselected":[{"key":"GSS_HEALTH","text":"Would you say your own health, in general, is excellent, good, fair, or poor?","model_mass":1,"human_mass":0.0007,"gap":0.9993},{"key":"GSS_CLASS","text":"If you were asked to use one of four names for your social class, which would you say you belong in: the lower class, the working class, the middle class, or the upper class?","model_mass":1,"human_mass":0.010781,"gap":0.989219},{"key":"GSS_TAX","text":"Do you consider the amount of federal income tax which you have to pay as too high, about right, or too low?","model_mass":1,"human_mass":0.025797,"gap":0.974203},{"key":"GSS_ABPOOR","text":"Please tell me whether or not you think it should be possible for a pregnant woman to obtain a legal abortion if the family has a very low income and cannot afford any more children?","model_mass":0.933317,"human_mass":0.039871,"gap":0.893446},{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373}]},{"provider":"Claude Sonnet 5.5","framework":"raw","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.209771,"mean_model_nonresponse_mass":0.199556,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.199556,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_GUNLAW","model_mass":1,"human_mass":0.03295,"gap":0.96705},{"key":"GSS_HEALTH","model_mass":0.9667,"human_mass":0.001396,"gap":0.965304},{"key":"GSS_NATFARE","model_mass":1,"human_mass":0.045323,"gap":0.954677},{"key":"GSS_NATARMS","model_mass":1,"human_mass":0.064099,"gap":0.935901},{"key":"GSS_CAPPUN","model_mass":1,"human_mass":0.068223,"gap":0.931777}],"top_option_overselected":[{"key":"GSS_GUNLAW","text":"Would you favor or oppose a law which would require a person to obtain a police permit before he or she could buy a gun?","model_mass":1,"human_mass":0.016324,"gap":0.983676},{"key":"GSS_NATFARE","text":"We are faced with many problems in this country, none of which can be solved easily or inexpensively. Are we spending too much, too little, or about the right amount on welfare?","model_mass":1,"human_mass":0.022374,"gap":0.977626},{"key":"GSS_NATARMS","text":"We are faced with many problems in this country, none of which can be solved easily or inexpensively. Are we spending too much, too little, or about the right amount on the military, armaments, and defense?","model_mass":1,"human_mass":0.031539,"gap":0.968461},{"key":"GSS_CAPPUN","text":"Do you favor or oppose the death penalty for persons convicted of murder?","model_mass":1,"human_mass":0.033495,"gap":0.966505},{"key":"GSS_ABANY","text":"Please tell me whether or not you think it should be possible for a pregnant woman to obtain a legal abortion if the woman wants it for any reason?","model_mass":1,"human_mass":0.033588,"gap":0.966412}]},{"provider":"Claude Haiku 4.5","framework":"raw","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.190269,"mean_model_nonresponse_mass":0.175761,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.109072,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_HEALTH","model_mass":1,"human_mass":0.001396,"gap":0.998604},{"key":"GSS_CLASS","model_mass":1,"human_mass":0.02169,"gap":0.97831},{"key":"GSS_TAX","model_mass":1,"human_mass":0.052342,"gap":0.947658},{"key":"GSS_ABANY","model_mass":1,"human_mass":0.068418,"gap":0.931582},{"key":"GSS_LETDIE1","model_mass":1,"human_mass":0.068945,"gap":0.931055}],"top_option_overselected":[{"key":"GSS_HEALTH","text":"Would you say your own health, in general, is excellent, good, fair, or poor?","model_mass":1,"human_mass":0.0007,"gap":0.9993},{"key":"GSS_CLASS","text":"If you were asked to use one of four names for your social class, which would you say you belong in: the lower class, the working class, the middle class, or the upper class?","model_mass":1,"human_mass":0.010781,"gap":0.989219},{"key":"GSS_TAX","text":"Do you consider the amount of federal income tax which you have to pay as too high, about right, or too low?","model_mass":1,"human_mass":0.025797,"gap":0.974203},{"key":"GSS_LETDIE1","text":"When a person has a disease that cannot be cured, do you think doctors should be allowed by law to end the patient's life by some painless means if the patient and his family request it?","model_mass":1,"human_mass":0.033869,"gap":0.966131},{"key":"GSS_GOD","text":"Which statement comes closest to expressing what you believe about God? I don't believe in God; I don't know whether there is a God and I don't believe there is any way to find out; I don't believe in a personal God, but I do believe in a Higher Power of some kind; I find myself believing in God some of the time, but not at others; while I have doubts, I feel that I do believe in God; or I know God really exists and I have no doubts about it.","model_mass":1,"human_mass":0.076527,"gap":0.923473}]},{"provider":"Althing (Sonnet 5.5)","framework":"product","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.17438,"mean_model_nonresponse_mass":0.174068,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.163742,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_CAPPUN","model_mass":1,"human_mass":0.068223,"gap":0.931777},{"key":"GSS_ABANY","model_mass":1,"human_mass":0.068418,"gap":0.931582},{"key":"GSS_ABPOOR","model_mass":1,"human_mass":0.081434,"gap":0.918566},{"key":"GSS_ABNOMORE","model_mass":1,"human_mass":0.08282,"gap":0.91718},{"key":"GSS_ABSINGLE","model_mass":0.900047,"human_mass":0.074724,"gap":0.825323}],"top_option_overselected":[{"key":"GSS_CAPPUN","text":"Do you favor or oppose the death penalty for persons convicted of murder?","model_mass":1,"human_mass":0.033495,"gap":0.966505},{"key":"GSS_ABNOMORE","text":"Please tell me whether or not you think it should be possible for a pregnant woman to obtain a legal abortion if she is married and does not want any more children?","model_mass":0.96,"human_mass":0.04051,"gap":0.91949},{"key":"GSS_GRASS","text":"Do you think the use of marijuana should be made legal or not?","model_mass":1,"human_mass":0.087191,"gap":0.912809},{"key":"GSS_ABANY","text":"Please tell me whether or not you think it should be possible for a pregnant woman to obtain a legal abortion if the woman wants it for any reason?","model_mass":0.933333,"human_mass":0.033588,"gap":0.899745},{"key":"GSS_ABPOOR","text":"Please tell me whether or not you think it should be possible for a pregnant woman to obtain a legal abortion if the family has a very low income and cannot afford any more children?","model_mass":0.933333,"human_mass":0.039871,"gap":0.893462}]},{"provider":"Althing (GPT-6 Luna)","framework":"product","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.15255,"mean_model_nonresponse_mass":0.156068,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.156068,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_FINRELA","model_mass":0.9655,"human_mass":0.014308,"gap":0.951192},{"key":"GSS_CLASS","model_mass":0.9667,"human_mass":0.02169,"gap":0.94501},{"key":"GSS_LETDIE1","model_mass":0.9667,"human_mass":0.068945,"gap":0.897755},{"key":"GSS_GUNLAW","model_mass":0.8667,"human_mass":0.03295,"gap":0.83375},{"key":"GSS_AFFRMACT","model_mass":0.8929,"human_mass":0.10282,"gap":0.79008}],"top_option_overselected":[{"key":"GSS_FINRELA","text":"Compared with American families in general, would you say your family income is far below average, below average, average, above average, or far above average?","model_mass":0.9655,"human_mass":0.007148,"gap":0.958352},{"key":"GSS_CLASS","text":"If you were asked to use one of four names for your social class, which would you say you belong in: the lower class, the working class, the middle class, or the upper class?","model_mass":0.9667,"human_mass":0.010781,"gap":0.955919},{"key":"GSS_LETDIE1","text":"When a person has a disease that cannot be cured, do you think doctors should be allowed by law to end the patient's life by some painless means if the patient and his family request it?","model_mass":0.9667,"human_mass":0.033869,"gap":0.932831},{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_GUNLAW","text":"Would you favor or oppose a law which would require a person to obtain a police permit before he or she could buy a gun?","model_mass":0.8667,"human_mass":0.016324,"gap":0.850376}]},{"provider":"Gemini 2.5 Flash","framework":"raw","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.123882,"mean_model_nonresponse_mass":0.097999,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.097999,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_FEPRESCH","model_mass":0.9333,"human_mass":0.021529,"gap":0.911771},{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555},{"key":"GSS_FINRELA","model_mass":0.6667,"human_mass":0.014308,"gap":0.652392},{"key":"GSS_GOD","model_mass":0.8,"human_mass":0.160041,"gap":0.639959},{"key":"GSS_NATRACE","model_mass":0.7,"human_mass":0.085912,"gap":0.614088}],"top_option_overselected":[{"key":"GSS_FEPRESCH","text":"Please indicate whether you strongly agree, agree, disagree, or strongly disagree with the following statement: A preschool child is likely to suffer if his or her mother works.","model_mass":0.9333,"human_mass":0.010683,"gap":0.922617},{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_GOD","text":"Which statement comes closest to expressing what you believe about God? I don't believe in God; I don't know whether there is a God and I don't believe there is any way to find out; I don't believe in a personal God, but I do believe in a Higher Power of some kind; I find myself believing in God some of the time, but not at others; while I have doubts, I feel that I do believe in God; or I know God really exists and I have no doubts about it.","model_mass":0.8,"human_mass":0.076527,"gap":0.723473},{"key":"GSS_FINRELA","text":"Compared with American families in general, would you say your family income is far below average, below average, average, above average, or far above average?","model_mass":0.6667,"human_mass":0.007148,"gap":0.659552},{"key":"GSS_NATRACE","text":"We are faced with many problems in this country, none of which can be solved easily or inexpensively. Are we spending too much, too little, or about the right amount on improving the conditions of Black people?","model_mass":0.7,"human_mass":0.041971,"gap":0.658029}]},{"provider":"Claude Sonnet 4.6","framework":"raw","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.12357,"mean_model_nonresponse_mass":0.089333,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.089333,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_HEALTH","model_mass":1,"human_mass":0.001396,"gap":0.998604},{"key":"GSS_CAPPUN","model_mass":1,"human_mass":0.068223,"gap":0.931777},{"key":"GSS_POLVIEWS","model_mass":1,"human_mass":0.073569,"gap":0.926431},{"key":"GSS_GOD","model_mass":1,"human_mass":0.160041,"gap":0.839959},{"key":"GSS_GRASS","model_mass":1,"human_mass":0.183657,"gap":0.816343}],"top_option_overselected":[{"key":"GSS_HEALTH","text":"Would you say your own health, in general, is excellent, good, fair, or poor?","model_mass":1,"human_mass":0.0007,"gap":0.9993},{"key":"GSS_CAPPUN","text":"Do you favor or oppose the death penalty for persons convicted of murder?","model_mass":1,"human_mass":0.033495,"gap":0.966505},{"key":"GSS_POLVIEWS","text":"I'm going to show you a seven-point scale on which the political views that people might hold are arranged from extremely liberal--point 1--to extremely conservative--point 7. Where would you place yourself on this scale?","model_mass":1,"human_mass":0.036095,"gap":0.963905},{"key":"GSS_GOD","text":"Which statement comes closest to expressing what you believe about God? I don't believe in God; I don't know whether there is a God and I don't believe there is any way to find out; I don't believe in a personal God, but I do believe in a Higher Power of some kind; I find myself believing in God some of the time, but not at others; while I have doubts, I feel that I do believe in God; or I know God really exists and I have no doubts about it.","model_mass":1,"human_mass":0.076527,"gap":0.923473},{"key":"GSS_GRASS","text":"Do you think the use of marijuana should be made legal or not?","model_mass":1,"human_mass":0.087191,"gap":0.912809}]},{"provider":"Althing (Haiku 4.5)","framework":"product","template":"structured","dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.094025,"mean_model_nonresponse_mass":0.064,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.064,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_LETDIE1","model_mass":1,"human_mass":0.068945,"gap":0.931055},{"key":"GSS_SUICIDE1","model_mass":1,"human_mass":0.080503,"gap":0.919497},{"key":"GSS_AFFRMACT","model_mass":0.9333,"human_mass":0.10282,"gap":0.83048},{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555},{"key":"GSS_CAPPUN","model_mass":0.5,"human_mass":0.068223,"gap":0.431777}],"top_option_overselected":[{"key":"GSS_LETDIE1","text":"When a person has a disease that cannot be cured, do you think doctors should be allowed by law to end the patient's life by some painless means if the patient and his family request it?","model_mass":1,"human_mass":0.033869,"gap":0.966131},{"key":"GSS_SUICIDE1","text":"Do you think a person has the right to end his or her own life if this person has an incurable disease?","model_mass":1,"human_mass":0.039411,"gap":0.960589},{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_AFFRMACT","text":"Some people say that because of past discrimination, Black people should be given preference in hiring and promotion. Others say that such preference in hiring and promotion of Black people is wrong because it discriminates against other Americans. What about your opinion--are you for or against preferential hiring and promotion of Black people, and do you feel that way strongly or not strongly?","model_mass":0.9333,"human_mass":0.050012,"gap":0.883288},{"key":"GSS_CAPPUN","text":"Do you favor or oppose the death penalty for persons convicted of murder?","model_mass":0.5,"human_mass":0.033495,"gap":0.466505}]},{"provider":"Althing (Haiku 4.5)","framework":"product","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.09165,"mean_model_nonresponse_mass":0.067222,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.066667,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_LETDIE1","model_mass":1,"human_mass":0.068945,"gap":0.931055},{"key":"GSS_SUICIDE1","model_mass":1,"human_mass":0.080503,"gap":0.919497},{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555},{"key":"GSS_GOD","model_mass":0.7,"human_mass":0.160041,"gap":0.539959},{"key":"GSS_GUNLAW","model_mass":0.4,"human_mass":0.03295,"gap":0.36705}],"top_option_overselected":[{"key":"GSS_LETDIE1","text":"When a person has a disease that cannot be cured, do you think doctors should be allowed by law to end the patient's life by some painless means if the patient and his family request it?","model_mass":1,"human_mass":0.033869,"gap":0.966131},{"key":"GSS_SUICIDE1","text":"Do you think a person has the right to end his or her own life if this person has an incurable disease?","model_mass":1,"human_mass":0.039411,"gap":0.960589},{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_GOD","text":"Which statement comes closest to expressing what you believe about God? I don't believe in God; I don't know whether there is a God and I don't believe there is any way to find out; I don't believe in a personal God, but I do believe in a Higher Power of some kind; I find myself believing in God some of the time, but not at others; while I have doubts, I feel that I do believe in God; or I know God really exists and I have no doubts about it.","model_mass":0.7,"human_mass":0.076527,"gap":0.623473},{"key":"GSS_GUNLAW","text":"Would you favor or oppose a law which would require a person to obtain a police permit before he or she could buy a gun?","model_mass":0.4,"human_mass":0.016324,"gap":0.383676}]},{"provider":"GPT-4o","framework":"raw","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.089945,"mean_model_nonresponse_mass":0.067111,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.067111,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_OWNGUN","model_mass":0.9,"human_mass":0.029577,"gap":0.870423},{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555},{"key":"GSS_FINRELA","model_mass":0.5667,"human_mass":0.014308,"gap":0.552392},{"key":"GSS_HEALTH","model_mass":0.5333,"human_mass":0.001396,"gap":0.531904},{"key":"GSS_AFFRMACT","model_mass":0.5,"human_mass":0.10282,"gap":0.39718}],"top_option_overselected":[{"key":"GSS_OWNGUN","text":"Do you happen to have in your home or garage any guns or revolvers?","model_mass":0.9,"human_mass":0.014678,"gap":0.885322},{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_FINRELA","text":"Compared with American families in general, would you say your family income is far below average, below average, average, above average, or far above average?","model_mass":0.5667,"human_mass":0.007148,"gap":0.559552},{"key":"GSS_HEALTH","text":"Would you say your own health, in general, is excellent, good, fair, or poor?","model_mass":0.5333,"human_mass":0.0007,"gap":0.5326},{"key":"GSS_AFFRMACT","text":"Some people say that because of past discrimination, Black people should be given preference in hiring and promotion. Others say that such preference in hiring and promotion of Black people is wrong because it discriminates against other Americans. What about your opinion--are you for or against preferential hiring and promotion of Black people, and do you feel that way strongly or not strongly?","model_mass":0.5,"human_mass":0.050012,"gap":0.449988}]},{"provider":"Althing (Haiku 5.5)","framework":"product","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.079941,"mean_model_nonresponse_mass":0.044023,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.038512,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_ABANY","model_mass":1,"human_mass":0.068418,"gap":0.931582},{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555},{"key":"GSS_NATFARE","model_mass":0.5667,"human_mass":0.045323,"gap":0.521377},{"key":"GSS_CONSCI","model_mass":0.333333,"human_mass":0.0418,"gap":0.291533},{"key":"GSS_CONCLERG","model_mass":0.119968,"human_mass":0.044295,"gap":0.075673}],"top_option_overselected":[{"key":"GSS_ABANY","text":"Please tell me whether or not you think it should be possible for a pregnant woman to obtain a legal abortion if the woman wants it for any reason?","model_mass":1,"human_mass":0.033588,"gap":0.966412},{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_NATFARE","text":"We are faced with many problems in this country, none of which can be solved easily or inexpensively. Are we spending too much, too little, or about the right amount on welfare?","model_mass":0.5667,"human_mass":0.022374,"gap":0.544326},{"key":"GSS_CONCLERG","text":"I am going to name an institution in this country. As far as the people running this institution are concerned, would you say you have a great deal of confidence, only some confidence, or hardly any confidence at all in organized religion?","model_mass":0.079968,"human_mass":0.021898,"gap":0.05807},{"key":"GSS_CONEDUC","text":"I am going to name an institution in this country. As far as the people running this institution are concerned, would you say you have a great deal of confidence, only some confidence, or hardly any confidence at all in education?","model_mass":0.0417,"human_mass":0.006063,"gap":0.035637}]},{"provider":"Gemini 3.5 Flash-Lite","framework":"raw","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.076215,"mean_model_nonresponse_mass":0.040537,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.040537,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_POLVIEWS","model_mass":1,"human_mass":0.073569,"gap":0.926431},{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555},{"key":"GSS_OWNGUN","model_mass":0.3667,"human_mass":0.029577,"gap":0.337123},{"key":"GSS_ABPOOR","model_mass":0.2,"human_mass":0.081434,"gap":0.118566},{"key":"GSS_HEALTH","model_mass":0.0667,"human_mass":0.001396,"gap":0.065304}],"top_option_overselected":[{"key":"GSS_POLVIEWS","text":"I'm going to show you a seven-point scale on which the political views that people might hold are arranged from extremely liberal--point 1--to extremely conservative--point 7. Where would you place yourself on this scale?","model_mass":1,"human_mass":0.036095,"gap":0.963905},{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_OWNGUN","text":"Do you happen to have in your home or garage any guns or revolvers?","model_mass":0.3667,"human_mass":0.014678,"gap":0.352022},{"key":"GSS_ABPOOR","text":"Please tell me whether or not you think it should be possible for a pregnant woman to obtain a legal abortion if the family has a very low income and cannot afford any more children?","model_mass":0.2,"human_mass":0.039871,"gap":0.160129},{"key":"GSS_GOD","text":"Which statement comes closest to expressing what you believe about God? I don't believe in God; I don't know whether there is a God and I don't believe there is any way to find out; I don't believe in a personal God, but I do believe in a Higher Power of some kind; I find myself believing in God some of the time, but not at others; while I have doubts, I feel that I do believe in God; or I know God really exists and I have no doubts about it.","model_mass":0.2069,"human_mass":0.076527,"gap":0.130373}]},{"provider":"Althing (GPT-6 Luna)","framework":"product","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.072654,"mean_model_nonresponse_mass":0.046281,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.046281,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_GOD","model_mass":1,"human_mass":0.160041,"gap":0.839959},{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555},{"key":"GSS_FINRELA","model_mass":0.4333,"human_mass":0.014308,"gap":0.418992},{"key":"GSS_CLASS","model_mass":0.2,"human_mass":0.02169,"gap":0.17831},{"key":"GSS_COURTS","model_mass":0.3667,"human_mass":0.277007,"gap":0.089693}],"top_option_overselected":[{"key":"GSS_GOD","text":"Which statement comes closest to expressing what you believe about God? I don't believe in God; I don't know whether there is a God and I don't believe there is any way to find out; I don't believe in a personal God, but I do believe in a Higher Power of some kind; I find myself believing in God some of the time, but not at others; while I have doubts, I feel that I do believe in God; or I know God really exists and I have no doubts about it.","model_mass":1,"human_mass":0.076527,"gap":0.923473},{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_FINRELA","text":"Compared with American families in general, would you say your family income is far below average, below average, average, above average, or far above average?","model_mass":0.4333,"human_mass":0.007148,"gap":0.426152},{"key":"GSS_COURTS","text":"In general, do you think the courts in this area deal too harshly or not harshly enough with criminals?","model_mass":0.3667,"human_mass":0.127287,"gap":0.239413},{"key":"GSS_CLASS","text":"If you were asked to use one of four names for your social class, which would you say you belong in: the lower class, the working class, the middle class, or the upper class?","model_mass":0.2,"human_mass":0.010781,"gap":0.189219}]},{"provider":"Althing (Gemini Flash Lite)","framework":"product","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.072509,"mean_model_nonresponse_mass":0.035112,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.035112,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555},{"key":"GSS_NATFARE","model_mass":0.4667,"human_mass":0.045323,"gap":0.421377},{"key":"GSS_SUICIDE1","model_mass":0.5,"human_mass":0.080503,"gap":0.419497},{"key":"GSS_NATSPAC","model_mass":0.2667,"human_mass":0.109812,"gap":0.156888},{"key":"GSS_NATAID","model_mass":0.2,"human_mass":0.061419,"gap":0.138581}],"top_option_overselected":[{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_SUICIDE1","text":"Do you think a person has the right to end his or her own life if this person has an incurable disease?","model_mass":0.5,"human_mass":0.039411,"gap":0.460589},{"key":"GSS_NATFARE","text":"We are faced with many problems in this country, none of which can be solved easily or inexpensively. Are we spending too much, too little, or about the right amount on welfare?","model_mass":0.4667,"human_mass":0.022374,"gap":0.444326},{"key":"GSS_NATSPAC","text":"We are faced with many problems in this country, none of which can be solved easily or inexpensively. Are we spending too much, too little, or about the right amount on the space exploration program?","model_mass":0.2667,"human_mass":0.053308,"gap":0.213392},{"key":"GSS_NATAID","text":"We are faced with many problems in this country, none of which can be solved easily or inexpensively. Are we spending too much, too little, or about the right amount on foreign aid?","model_mass":0.2,"human_mass":0.030227,"gap":0.169773}]},{"provider":"Althing (Muse Glimmer 30B)","framework":"product","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.069628,"mean_model_nonresponse_mass":0.03386,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.033327,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555},{"key":"GSS_LETDIE1","model_mass":0.5667,"human_mass":0.068945,"gap":0.497755},{"key":"GSS_NATCRIME","model_mass":0.2667,"human_mass":0.039162,"gap":0.227538},{"key":"GSS_GOD","model_mass":0.3462,"human_mass":0.160041,"gap":0.186159},{"key":"GSS_CONMEDIC","model_mass":0.16,"human_mass":0.010545,"gap":0.149455}],"top_option_overselected":[{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_LETDIE1","text":"When a person has a disease that cannot be cured, do you think doctors should be allowed by law to end the patient's life by some painless means if the patient and his family request it?","model_mass":0.5667,"human_mass":0.033869,"gap":0.532831},{"key":"GSS_GOD","text":"Which statement comes closest to expressing what you believe about God? I don't believe in God; I don't know whether there is a God and I don't believe there is any way to find out; I don't believe in a personal God, but I do believe in a Higher Power of some kind; I find myself believing in God some of the time, but not at others; while I have doubts, I feel that I do believe in God; or I know God really exists and I have no doubts about it.","model_mass":0.3462,"human_mass":0.076527,"gap":0.269673},{"key":"GSS_NATCRIME","text":"We are faced with many problems in this country, none of which can be solved easily or inexpensively. Are we spending too much, too little, or about the right amount on halting the rising crime rate?","model_mass":0.2667,"human_mass":0.019409,"gap":0.247291},{"key":"GSS_CONMEDIC","text":"I am going to name an institution in this country. As far as the people running this institution are concerned, would you say you have a great deal of confidence, only some confidence, or hardly any confidence at all in medicine?","model_mass":0.12,"human_mass":0.005272,"gap":0.114728}]},{"provider":"Gemini Flash Lite","framework":"raw","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.069609,"mean_model_nonresponse_mass":0.036444,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.036,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555},{"key":"GSS_AFFRMACT","model_mass":0.7,"human_mass":0.10282,"gap":0.59718},{"key":"GSS_NATFARE","model_mass":0.3,"human_mass":0.045323,"gap":0.254677},{"key":"GSS_GOD","model_mass":0.3334,"human_mass":0.160041,"gap":0.173359},{"key":"GSS_FEPRESCH","model_mass":0.1,"human_mass":0.021529,"gap":0.078471}],"top_option_overselected":[{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_AFFRMACT","text":"Some people say that because of past discrimination, Black people should be given preference in hiring and promotion. Others say that such preference in hiring and promotion of Black people is wrong because it discriminates against other Americans. What about your opinion--are you for or against preferential hiring and promotion of Black people, and do you feel that way strongly or not strongly?","model_mass":0.7,"human_mass":0.050012,"gap":0.649988},{"key":"GSS_NATFARE","text":"We are faced with many problems in this country, none of which can be solved easily or inexpensively. Are we spending too much, too little, or about the right amount on welfare?","model_mass":0.3,"human_mass":0.022374,"gap":0.277626},{"key":"GSS_GOD","text":"Which statement comes closest to expressing what you believe about God? I don't believe in God; I don't know whether there is a God and I don't believe there is any way to find out; I don't believe in a personal God, but I do believe in a Higher Power of some kind; I find myself believing in God some of the time, but not at others; while I have doubts, I feel that I do believe in God; or I know God really exists and I have no doubts about it.","model_mass":0.3334,"human_mass":0.076527,"gap":0.256873},{"key":"GSS_FEPRESCH","text":"Please indicate whether you strongly agree, agree, disagree, or strongly disagree with the following statement: A preschool child is likely to suffer if his or her mother works.","model_mass":0.1,"human_mass":0.010683,"gap":0.089317}]},{"provider":"GPT-4o-mini","framework":"raw","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.068668,"mean_model_nonresponse_mass":0.025333,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.025333,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555},{"key":"GSS_GOD","model_mass":0.9,"human_mass":0.160041,"gap":0.739959}],"top_option_overselected":[{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_GOD","text":"Which statement comes closest to expressing what you believe about God? I don't believe in God; I don't know whether there is a God and I don't believe there is any way to find out; I don't believe in a personal God, but I do believe in a Higher Power of some kind; I find myself believing in God some of the time, but not at others; while I have doubts, I feel that I do believe in God; or I know God really exists and I have no doubts about it.","model_mass":0.9,"human_mass":0.076527,"gap":0.823473}]},{"provider":"DeepSeek V4.1 Flash","framework":"raw","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.068222,"mean_model_nonresponse_mass":0.026667,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.026667,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_GOD","model_mass":0.9333,"human_mass":0.160041,"gap":0.773259},{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555}],"top_option_overselected":[{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_GOD","text":"Which statement comes closest to expressing what you believe about God? I don't believe in God; I don't know whether there is a God and I don't believe there is any way to find out; I don't believe in a personal God, but I do believe in a Higher Power of some kind; I find myself believing in God some of the time, but not at others; while I have doubts, I feel that I do believe in God; or I know God really exists and I have no doubts about it.","model_mass":0.9333,"human_mass":0.076527,"gap":0.856773},{"key":"GSS_AFFRMACT","text":"Some people say that because of past discrimination, Black people should be given preference in hiring and promotion. Others say that such preference in hiring and promotion of Black people is wrong because it discriminates against other Americans. What about your opinion--are you for or against preferential hiring and promotion of Black people, and do you feel that way strongly or not strongly?","model_mass":0.0667,"human_mass":0.050012,"gap":0.016688}]},{"provider":"Qwen 3.8 Flash","framework":"raw","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.063177,"mean_model_nonresponse_mass":0.028444,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.028444,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555},{"key":"GSS_FINRELA","model_mass":0.2333,"human_mass":0.014308,"gap":0.218992},{"key":"GSS_CAPPUN","model_mass":0.2333,"human_mass":0.068223,"gap":0.165077},{"key":"GSS_GOD","model_mass":0.2667,"human_mass":0.160041,"gap":0.106659},{"key":"GSS_NATAID","model_mass":0.1667,"human_mass":0.061419,"gap":0.105281}],"top_option_overselected":[{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_FINRELA","text":"Compared with American families in general, would you say your family income is far below average, below average, average, above average, or far above average?","model_mass":0.2333,"human_mass":0.007148,"gap":0.226152},{"key":"GSS_CAPPUN","text":"Do you favor or oppose the death penalty for persons convicted of murder?","model_mass":0.2333,"human_mass":0.033495,"gap":0.199805},{"key":"GSS_GOD","text":"Which statement comes closest to expressing what you believe about God? I don't believe in God; I don't know whether there is a God and I don't believe there is any way to find out; I don't believe in a personal God, but I do believe in a Higher Power of some kind; I find myself believing in God some of the time, but not at others; while I have doubts, I feel that I do believe in God; or I know God really exists and I have no doubts about it.","model_mass":0.2667,"human_mass":0.076527,"gap":0.190173},{"key":"GSS_NATAID","text":"We are faced with many problems in this country, none of which can be solved easily or inexpensively. Are we spending too much, too little, or about the right amount on foreign aid?","model_mass":0.1667,"human_mass":0.030227,"gap":0.136473}]},{"provider":"Althing (DeepSeek V4.1 Flash)","framework":"product","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.0621,"mean_model_nonresponse_mass":0.018765,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.018765,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555},{"key":"GSS_GOD","model_mass":0.4074,"human_mass":0.160041,"gap":0.247359}],"top_option_overselected":[{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_GOD","text":"Which statement comes closest to expressing what you believe about God? I don't believe in God; I don't know whether there is a God and I don't believe there is any way to find out; I don't believe in a personal God, but I do believe in a Higher Power of some kind; I find myself believing in God some of the time, but not at others; while I have doubts, I feel that I do believe in God; or I know God really exists and I have no doubts about it.","model_mass":0.4074,"human_mass":0.076527,"gap":0.330873}]},{"provider":"Althing (Sonnet 4.6)","framework":"product","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.060936,"mean_model_nonresponse_mass":0.013333,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.013333,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555}],"top_option_overselected":[{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373}]},{"provider":"Althing (Gemini 3.5 Flash-Lite)","framework":"product","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.060936,"mean_model_nonresponse_mass":0.013333,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.013333,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555}],"top_option_overselected":[{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373}]},{"provider":"Llama 3.3 70B","framework":"raw","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.060554,"mean_model_nonresponse_mass":0.014221,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.014221,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555},{"key":"GSS_FINRELA","model_mass":0.0333,"human_mass":0.014308,"gap":0.018992}],"top_option_overselected":[{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_FINRELA","text":"Compared with American families in general, would you say your family income is far below average, below average, average, above average, or far above average?","model_mass":0.0333,"human_mass":0.007148,"gap":0.026152}]},{"provider":"Althing (Sonnet 4)","framework":"product","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.059346,"mean_model_nonresponse_mass":0.015189,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.015189,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555},{"key":"GSS_NATAID","model_mass":0.0714,"human_mass":0.061419,"gap":0.009981}],"top_option_overselected":[{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_NATAID","text":"We are faced with many problems in this country, none of which can be solved easily or inexpensively. Are we spending too much, too little, or about the right amount on foreign aid?","model_mass":0.0714,"human_mass":0.030227,"gap":0.041173}]},{"provider":"Althing (GPT-4o-mini)","framework":"product","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.059334,"mean_model_nonresponse_mass":0.016,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.016,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555},{"key":"GSS_GOD","model_mass":0.2,"human_mass":0.160041,"gap":0.039959}],"top_option_overselected":[{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_GOD","text":"Which statement comes closest to expressing what you believe about God? I don't believe in God; I don't know whether there is a God and I don't believe there is any way to find out; I don't believe in a personal God, but I do believe in a Higher Power of some kind; I find myself believing in God some of the time, but not at others; while I have doubts, I feel that I do believe in God; or I know God really exists and I have no doubts about it.","model_mass":0.2,"human_mass":0.076527,"gap":0.123473}]},{"provider":"Althing (Qwen 3.8 Flash)","framework":"product","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.055907,"mean_model_nonresponse_mass":0.031041,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.031041,"mean_human_option_mass":0.026358,"top_overselected":[{"key":"GSS_POSTLIFE","model_mass":1,"human_mass":0.251445,"gap":0.748555},{"key":"GSS_GOD","model_mass":0.3448,"human_mass":0.160041,"gap":0.184759},{"key":"GSS_SUICIDE1","model_mass":0.2414,"human_mass":0.080503,"gap":0.160897},{"key":"GSS_NATSPAC","model_mass":0.2,"human_mass":0.109812,"gap":0.090188},{"key":"GSS_GUNLAW","model_mass":0.0714,"human_mass":0.03295,"gap":0.03845}],"top_option_overselected":[{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":1,"human_mass":0.116627,"gap":0.883373},{"key":"GSS_GOD","text":"Which statement comes closest to expressing what you believe about God? I don't believe in God; I don't know whether there is a God and I don't believe there is any way to find out; I don't believe in a personal God, but I do believe in a Higher Power of some kind; I find myself believing in God some of the time, but not at others; while I have doubts, I feel that I do believe in God; or I know God really exists and I have no doubts about it.","model_mass":0.3448,"human_mass":0.076527,"gap":0.268273},{"key":"GSS_SUICIDE1","text":"Do you think a person has the right to end his or her own life if this person has an incurable disease?","model_mass":0.2414,"human_mass":0.039411,"gap":0.201989},{"key":"GSS_NATSPAC","text":"We are faced with many problems in this country, none of which can be solved easily or inexpensively. Are we spending too much, too little, or about the right amount on the space exploration program?","model_mass":0.2,"human_mass":0.053308,"gap":0.146692},{"key":"GSS_GUNLAW","text":"Would you favor or oppose a law which would require a person to obtain a police permit before he or she could buy a gun?","model_mass":0.0714,"human_mass":0.016324,"gap":0.055076}]},{"provider":"Althing (Llama 3.3 70B)","framework":"product","template":null,"dataset":"gss","n_questions":75,"mean_abs_nonresponse_gap":0.052085,"mean_model_nonresponse_mass":0.002223,"mean_human_nonresponse_mass":0.054307,"mean_model_option_mass":0.002223,"mean_human_option_mass":0.026358,"top_overselected":[],"top_option_overselected":[{"key":"GSS_POSTLIFE","text":"Do you believe there is a life after death?","model_mass":0.1667,"human_mass":0.116627,"gap":0.050073}]}],"sensitive_topic_sidestepping":{"headline":"Safety-aligned models over-select explicit nonresponse options on sensitive items","provider":"Muse Glimmer 30B","dataset":"gss","n_questions":75,"mean_model_nonresponse_mass":0.738128,"mean_human_nonresponse_mass":0.026358,"items":[{"key":"GSS_CONFINAN","text":"I am going to name an institution in this country. As far as the people running this institution are concerned, would you say you have a great deal of confidence, only some confidence, or hardly any confidence at all in banks and financial institutions?","model_mass":1,"human_mass":0.006557,"gap":0.993443},{"key":"GSS_XMARSEX","text":"What is your opinion about a married person having sexual relations with someone other than the marriage partner--is it always wrong, almost always wrong, wrong only sometimes, or not wrong at all?","model_mass":1,"human_mass":0.008034,"gap":0.991966},{"key":"GSS_PORNLAW","text":"Which of these statements comes closest to your feelings about pornography laws? There should be laws against the distribution of pornography whatever the age; there should be laws against the distribution of pornography to persons under 18; or there should be no laws forbidding the distribution of pornography.","model_mass":1,"human_mass":0.008919,"gap":0.991081},{"key":"GSS_CLASS","text":"If you were asked to use one of four names for your social class, which would you say you belong in: the lower class, the working class, the middle class, or the upper class?","model_mass":1,"human_mass":0.010781,"gap":0.989219},{"key":"GSS_CONJUDGE","text":"I am going to name an institution in this country. As far as the people running this institution are concerned, would you say you have a great deal of confidence, only some confidence, or hardly any confidence at all in the U.S. Supreme Court?","model_mass":1,"human_mass":0.01429,"gap":0.98571}],"note":"Share of answers that pick a listed 'don't know'-style option, excluding prose refusals. The model selects a legitimate option rather than refusing, so refusal-rate metrics miss it and it only shows in the option distribution."},"lever_hierarchy":[{"name":"Ensemble blending","effect_min":-0.2,"effect_max":7.1,"cost":"zero","status":"done"},{"name":"Per-model optimal temperature","effect_min":1.5,"effect_max":4.5,"cost":"low","status":"actionable"},{"name":"Demographic conditioning","effect_min":4,"effect_max":10.5,"cost":"moderate","status":"scientific"},{"name":"Persona template","effect_min":0,"effect_max":0,"cost":"zero","status":"done","note":"'current' template already optimal (+5.8 pts over the best alternative); no further gain available from this lever."}],"comparison_sets":{"ensemble_comparison":"best_single = highest recomputed-SPS non-ensemble, non-baseline leaderboard row (raw or product framework) evaluated on at least as many questions as the ensemble, after per-(model, framework, dataset) dedup; random_baseline = recomputed SPS of the random-baseline run for the same dataset (n in random_baseline_n).","ensemble_error_correlation":"For each dataset's published ensemble: pairwise Pearson r between the constituent runs' committed per-question JSD vectors (each run's error magnitude vs the canonical human distribution), over the ensemble's common question set, read from the exact files named in ensemble_sources. Measures whether constituents err on the same questions. Signed per-option residual correlations need the stripped human_distribution fields (#308) and are carried as the ensemble_signed_error_correlation asserted constant.","temperature_sweep":"OpinionsQA sweep-tagged Althing runs (' t=' provider suffix); mean/std of recomputed SPS across the n_runs replications per (model, temperature) cell.","conditioning_results":"Per-group conditioned SubPOP evaluations (Haiku 4.5, t=0.85); mean p_dist / p_cond across n_replications runs carrying a demographic_breakdown for the group.","template_comparison":"SubPOP persona-template variant runs (Haiku 4.5, t=0.85, 100 questions); mean/std of recomputed SPS across n_runs.","elicitation_comparison":"Matched pairs of deduped runs differing only in elicitation mode: a run whose config carries an explicit elicitation template variant (e.g. tpl=structured, schema-forced capture) vs the natural-elicitation run of the identical configuration (same base provider, dataset, samples_per_question, question_set_hash, temperature, effort, and question count). All scores recomputed from per-question rows; deltas are structured minus natural.","nonresponse_fidelity":"Per deduped non-baseline run whose committed per-question rows still carry human_distribution (full-tier datasets; gated files are stripped per #308): mean |model explicit-nonresponse mass (DK-style option mass + parsed refusal rate) - human's|, plus the five items with the largest model over-selection."},"asserted_constants":[{"name":"blend_weighting_equivalence","value":"equal-weight ~= score-proportional ~= inverse-JSD (to 3 decimals)","source":"Session-1 ensemble weighting comparison (2026-04-12); the alternative-weighting blend files were not committed to leaderboard-results/, so this is not reproducible from artifacts."},{"name":"per_question_improvement_share","value":"72-81% of individual questions improve under blending","source":"Session-1 per-question blend analysis (2026-04-12), computed under the retired parity-2 convention; not recomputed since."},{"name":"oracle_ceiling_gap","value":"per-question oracle model selection barely exceeds the equal blend","source":"Session-1 oracle analysis (2026-04-12); oracle blend files were not committed to leaderboard-results/."},{"name":"base_entropy_h5","value":"base output entropy (bits, default temperature): GPT-4o-mini 0.22, Claude Haiku 4.5 0.36, Gemini Flash Lite 0.56","source":"Experiment H5 notebook analysis (2026-04-12) of default-temperature model_distribution rows; the notebook output was not committed, so the exact values are asserted."},{"name":"ensemble_signed_error_correlation","value":"pairwise Pearson r of constituents' per-option signed residuals (model probability − human probability) on the ensemble common question sets: globalopinionqa 0.27–0.44, opinionsqa 0.36–0.38, subpop 0.36–0.42 — all pairs moderately positively correlated","source":"Computed 2026-07-16 from the committed constituents' per-question model_distribution rows plus canonical human-distribution rehydration (synthbench.human_distributions); committed gated artifacts strip human_distribution (#308), so the signed residuals are not derivable from public artifacts alone. The JSD-based correlations in ensemble_error_correlation ARE derivable and are recomputed by the drift guard."},{"name":"template_refusal_collapse","value":"P_refuse collapses from ~0.80 to 0.40-0.50 on templates with unfilled format-string placeholders","source":"Session-1 template-variant analysis (2026-04-12); per-component P_refuse for the template runs is derivable in principle but the published block only tracks composite SPS for templates."}],"caveats":[]},"concordance":{"globalopinionqa":{"models":["Claude Haiku 4.5","Claude Haiku 5.5","Claude Sonnet 4","Claude Sonnet 4.6","Claude Sonnet 5.5","DeepSeek V4.1 Flash","GPT-4o","GPT-4o-mini","GPT-6 Luna","Gemini 2.5 Flash","Gemini 3.5 Flash-Lite","Gemini 3.8 Flash","Gemini Flash Lite","Llama 3.3 70B","Muse Glimmer 30B","Qwen 3.8 Flash"],"matrix":[[0,0.246159,0.180343,0.279114,0.286459,0.255402,0.238816,0.297846,0.198614,0.303834,0.213113,0.356841,0.266883,0.255543,0.349853,0.194352],[0.246159,0,0.26449,0.351858,0.29983,0.254444,0.279014,0.316861,0.225671,0.361186,0.260172,0.345835,0.344536,0.347307,0.365967,0.251515],[0.180343,0.26449,0,0.274853,0.305166,0.293965,0.266692,0.332837,0.176281,0.319974,0.295247,0.340507,0.276184,0.342849,0.32098,0.243091],[0.279114,0.351858,0.274853,0,0.302767,0.288786,0.323435,0.355235,0.217146,0.363608,0.282863,0.307931,0.338152,0.306379,0.330162,0.264019],[0.286459,0.29983,0.305166,0.302767,0,0.243534,0.319058,0.285458,0.260374,0.388585,0.260908,0.327279,0.342103,0.311946,0.390278,0.217645],[0.255402,0.254444,0.293965,0.288786,0.243534,0,0.228786,0.221823,0.235579,0.275324,0.247623,0.332337,0.265035,0.223506,0.338354,0.162421],[0.238816,0.279014,0.266692,0.323435,0.319058,0.228786,0,0.17394,0.196646,0.175182,0.225868,0.321922,0.197259,0.23565,0.365283,0.183381],[0.297846,0.316861,0.332837,0.355235,0.285458,0.221823,0.17394,0,0.276161,0.275595,0.290464,0.390499,0.247532,0.208468,0.413751,0.152269],[0.198614,0.225671,0.176281,0.217146,0.260374,0.235579,0.196646,0.276161,0,0.285073,0.222728,0.247352,0.246203,0.308911,0.21838,0.204659],[0.303834,0.361186,0.319974,0.363608,0.388585,0.275324,0.175182,0.275595,0.285073,0,0.27003,0.417608,0.201592,0.299732,0.458111,0.243859],[0.213113,0.260172,0.295247,0.282863,0.260908,0.247623,0.225868,0.290464,0.222728,0.27003,0,0.336506,0.238534,0.260617,0.369518,0.214527],[0.356841,0.345835,0.340507,0.307931,0.327279,0.332337,0.321922,0.390499,0.247352,0.417608,0.336506,0,0.381072,0.427143,0.146238,0.352204],[0.266883,0.344536,0.276184,0.338152,0.342103,0.265035,0.197259,0.247532,0.246203,0.201592,0.238534,0.381072,0,0.284287,0.414148,0.204452],[0.255543,0.347307,0.342849,0.306379,0.311946,0.223506,0.23565,0.208468,0.308911,0.299732,0.260617,0.427143,0.284287,0,0.440143,0.185388],[0.349853,0.365967,0.32098,0.330162,0.390278,0.338354,0.365283,0.413751,0.21838,0.458111,0.369518,0.146238,0.414148,0.440143,0,0.352422],[0.194352,0.251515,0.243091,0.264019,0.217645,0.162421,0.183381,0.152269,0.204659,0.243859,0.214527,0.352204,0.204452,0.185388,0.352422,0]],"mean_cross_model_jsd":0.284169,"mean_human_jsd":0.353224},"gss":{"models":["Claude Haiku 4.5","Claude Haiku 5.5","Claude Sonnet 4","Claude Sonnet 4.6","Claude Sonnet 5.5","DeepSeek V4.1 Flash","GPT-4o","GPT-4o-mini","GPT-6 Luna","Gemini 2.5 Flash","Gemini 3.5 Flash-Lite","Gemini 3.8 Flash","Gemini Flash Lite","Llama 3.3 70B","Muse Glimmer 30B","Qwen 3.8 Flash"],"matrix":[[0,0.241424,0.159388,0.231797,0.271744,0.215321,0.249453,0.261596,0.265208,0.321019,0.250109,0.606912,0.28325,0.234589,0.573304,0.238164],[0.241424,0,0.262434,0.258403,0.247462,0.271241,0.282421,0.272212,0.257139,0.320818,0.292083,0.504086,0.354581,0.273937,0.508759,0.267186],[0.159388,0.262434,0,0.278308,0.288718,0.284003,0.264449,0.333462,0.283385,0.332578,0.296503,0.533438,0.334878,0.273724,0.54067,0.275929],[0.231797,0.258403,0.278308,0,0.323304,0.230757,0.183975,0.256953,0.278004,0.241054,0.184839,0.630568,0.251032,0.22253,0.636527,0.197141],[0.271744,0.247462,0.288718,0.323304,0,0.271286,0.367116,0.342169,0.3454,0.416989,0.335242,0.541892,0.406719,0.313548,0.52827,0.28524],[0.215321,0.271241,0.284003,0.230757,0.271286,0,0.213357,0.177187,0.344276,0.226638,0.157614,0.728691,0.216007,0.12398,0.69139,0.086958],[0.249453,0.282421,0.264449,0.183975,0.367116,0.213357,0,0.215248,0.279493,0.168819,0.193576,0.649413,0.18825,0.158037,0.641643,0.180729],[0.261596,0.272212,0.333462,0.256953,0.342169,0.177187,0.215248,0,0.366866,0.26154,0.232001,0.711818,0.216246,0.174352,0.726645,0.165228],[0.265208,0.257139,0.283385,0.278004,0.3454,0.344276,0.279493,0.366866,0,0.318809,0.377607,0.404434,0.325985,0.371441,0.348921,0.303532],[0.321019,0.320818,0.332578,0.241054,0.416989,0.226638,0.168819,0.26154,0.318809,0,0.189227,0.629288,0.157055,0.221162,0.60326,0.199472],[0.250109,0.292083,0.296503,0.184839,0.335242,0.157614,0.193576,0.232001,0.377607,0.189227,0,0.666659,0.238753,0.16508,0.653359,0.15202],[0.606912,0.504086,0.533438,0.630568,0.541892,0.728691,0.649413,0.711818,0.404434,0.629288,0.666659,0,0.712333,0.69128,0.135879,0.686928],[0.28325,0.354581,0.334878,0.251032,0.406719,0.216007,0.18825,0.216246,0.325985,0.157055,0.238753,0.712333,0,0.229723,0.681211,0.192591],[0.234589,0.273937,0.273724,0.22253,0.313548,0.12398,0.158037,0.174352,0.371441,0.221162,0.16508,0.69128,0.229723,0,0.687167,0.147614],[0.573304,0.508759,0.54067,0.636527,0.52827,0.69139,0.641643,0.726645,0.348921,0.60326,0.653359,0.135879,0.681211,0.687167,0,0.662337],[0.238164,0.267186,0.275929,0.197141,0.28524,0.086958,0.180729,0.165228,0.303532,0.199472,0.15202,0.686928,0.192591,0.147614,0.662337,0]],"mean_cross_model_jsd":0.337798,"mean_human_jsd":0.380349},"opinionsqa":{"models":["Claude Haiku 4.5","Claude Haiku 5.5","Claude Sonnet 4","Claude Sonnet 4.6","Claude Sonnet 5.5","DeepSeek V4.1 Flash","GPT-4o","GPT-4o-mini","GPT-6 Luna","Gemini 2.5 Flash","Gemini 3.5 Flash-Lite","Gemini 3.8 Flash","Gemini Flash Lite","Llama 3.3 70B","Muse Glimmer 30B","Qwen 3.8 Flash"],"matrix":[[0,0.307645,0.2855,0.221357,0.239215,0.262031,0.293193,0.347915,0.322021,0.425474,0.290285,0.39022,0.353172,0.272842,0.360835,0.212191],[0.307645,0,0.350481,0.382021,0.31586,0.284125,0.339933,0.314911,0.253814,0.413302,0.274201,0.389874,0.3991,0.300432,0.315605,0.206025],[0.2855,0.350481,0,0.346336,0.326902,0.319842,0.302924,0.455507,0.25123,0.452665,0.332402,0.330635,0.456552,0.33791,0.309532,0.295602],[0.221357,0.382021,0.346336,0,0.254789,0.236227,0.326143,0.379561,0.431085,0.421504,0.272014,0.40935,0.413781,0.279939,0.475404,0.268844],[0.239215,0.31586,0.326902,0.254789,0,0.272409,0.384176,0.409002,0.390329,0.499067,0.336913,0.330447,0.454805,0.309498,0.360478,0.233663],[0.262031,0.284125,0.319842,0.236227,0.272409,0,0.212574,0.225326,0.350814,0.322829,0.197357,0.429453,0.307993,0.168045,0.430908,0.157259],[0.293193,0.339933,0.302924,0.326143,0.384176,0.212574,0,0.173246,0.309297,0.228101,0.249285,0.435487,0.217227,0.243862,0.424266,0.225055],[0.347915,0.314911,0.455507,0.379561,0.409002,0.225326,0.173246,0,0.406082,0.257222,0.279376,0.514235,0.194652,0.272958,0.495786,0.21315],[0.322021,0.253814,0.25123,0.431085,0.390329,0.350814,0.309297,0.406082,0,0.457195,0.404014,0.396216,0.426969,0.366341,0.291141,0.308328],[0.425474,0.413302,0.452665,0.421504,0.499067,0.322829,0.228101,0.257222,0.457195,0,0.290853,0.541632,0.222285,0.358983,0.541452,0.311497],[0.290285,0.274201,0.332402,0.272014,0.336913,0.197357,0.249285,0.279376,0.404014,0.290853,0,0.43154,0.311272,0.297874,0.413998,0.242445],[0.39022,0.389874,0.330635,0.40935,0.330447,0.429453,0.435487,0.514235,0.396216,0.541632,0.43154,0,0.551818,0.413522,0.253886,0.396923],[0.353172,0.3991,0.456552,0.413781,0.454805,0.307993,0.217227,0.194652,0.426969,0.222285,0.311272,0.551818,0,0.343414,0.520075,0.263588],[0.272842,0.300432,0.33791,0.279939,0.309498,0.168045,0.243862,0.272958,0.366341,0.358983,0.297874,0.413522,0.343414,0,0.421414,0.228086],[0.360835,0.315605,0.309532,0.475404,0.360478,0.430908,0.424266,0.495786,0.291141,0.541452,0.413998,0.253886,0.520075,0.421414,0,0.368758],[0.212191,0.206025,0.295602,0.268844,0.233663,0.157259,0.225055,0.21315,0.308328,0.311497,0.242445,0.396923,0.263588,0.228086,0.368758,0]],"mean_cross_model_jsd":0.334237,"mean_human_jsd":0.319334},"subpop":{"models":["Claude Haiku 4.5","Claude Haiku 5.5","Claude Sonnet 4","Claude Sonnet 4.6","Claude Sonnet 5.5","DeepSeek V4.1 Flash","GPT-4o","GPT-4o-mini","GPT-6 Luna","Gemini 2.5 Flash","Gemini 3.5 Flash-Lite","Gemini 3.8 Flash","Gemini Flash Lite","Llama 3.3 70B","Muse Glimmer 30B","Qwen 3.8 Flash"],"matrix":[[0,0.31944,0.236089,0.222085,0.273225,0.276195,0.344741,0.402044,0.238521,0.280007,0.332655,0.577206,0.299913,0.347801,0.47595,0.253476],[0.31944,0,0.246473,0.268055,0.263016,0.370055,0.345304,0.472773,0.214521,0.354955,0.350306,0.490969,0.373672,0.426086,0.425961,0.335354],[0.236089,0.246473,0,0.239457,0.297887,0.349949,0.360026,0.507359,0.20094,0.384146,0.395297,0.521926,0.402354,0.412239,0.378501,0.31692],[0.222085,0.268055,0.239457,0,0.283124,0.296369,0.367902,0.469304,0.267794,0.310777,0.291942,0.592844,0.350796,0.332143,0.512504,0.291163],[0.273225,0.263016,0.297887,0.283124,0,0.301268,0.39576,0.471151,0.277021,0.343659,0.351754,0.426752,0.368278,0.370533,0.396653,0.293577],[0.276195,0.370055,0.349949,0.296369,0.301268,0,0.294876,0.312355,0.270444,0.281286,0.235972,0.623798,0.303745,0.166422,0.556895,0.161345],[0.344741,0.345304,0.360026,0.367902,0.39576,0.294876,0,0.197668,0.232544,0.243647,0.353517,0.562365,0.261831,0.323444,0.501608,0.225996],[0.402044,0.472773,0.507359,0.469304,0.471151,0.312355,0.197668,0,0.40071,0.309591,0.377414,0.691493,0.250331,0.312707,0.657988,0.231547],[0.238521,0.214521,0.20094,0.267794,0.277021,0.270444,0.232544,0.40071,0,0.312897,0.335463,0.471826,0.334326,0.349725,0.324026,0.28063],[0.280007,0.354955,0.384146,0.310777,0.343659,0.281286,0.243647,0.309591,0.312897,0,0.279831,0.564057,0.183923,0.30145,0.523058,0.205398],[0.332655,0.350306,0.395297,0.291942,0.351754,0.235972,0.353517,0.377414,0.335463,0.279831,0,0.628238,0.291516,0.303856,0.573355,0.272238],[0.577206,0.490969,0.521926,0.592844,0.426752,0.623798,0.562365,0.691493,0.471826,0.564057,0.628238,0,0.607751,0.669761,0.157686,0.614258],[0.299913,0.373672,0.402354,0.350796,0.368278,0.303745,0.261831,0.250331,0.334326,0.183923,0.291516,0.607751,0,0.364411,0.562173,0.230654],[0.347801,0.426086,0.412239,0.332143,0.370533,0.166422,0.323444,0.312707,0.349725,0.30145,0.303856,0.669761,0.364411,0,0.603174,0.224015],[0.47595,0.425961,0.378501,0.512504,0.396653,0.556895,0.501608,0.657988,0.324026,0.523058,0.573355,0.157686,0.562173,0.603174,0,0.546399],[0.253476,0.335354,0.31692,0.291163,0.293577,0.161345,0.225996,0.231547,0.28063,0.205398,0.272238,0.614258,0.230654,0.224015,0.546399,0]],"mean_cross_model_jsd":0.361705,"mean_human_jsd":0.391038}},"base":"/","slugs":{"ensemble-complementary-4":"ensemble-complementary-4","ensemble-complementary-top":"ensemble-complementary-top","ensemble-flash3-raw":"ensemble-flash3-raw","ensemble-complementary-mid":"ensemble-complementary-mid","qwen/qwen3.8-flash":"qwen-qwen3-8-flash","deepseek/deepseek-v4.1-flash":"deepseek-deepseek-v4-1-flash","ensemble":"ensemble","anthropic/claude-haiku-4-5":"anthropic-claude-haiku-4-5","google/gemini-3.5-flash-lite":"google-gemini-3-5-flash-lite","anthropic/claude-sonnet-4":"anthropic-claude-sonnet-4","meta/muse-glimmer-30b":"meta-muse-glimmer-30b","anthropic/claude-haiku-5-5":"anthropic-claude-haiku-5-5","openai/gpt-6-luna":"openai-gpt-6-luna","google/gemini-2.5-flash-lite":"google-gemini-2-5-flash-lite","anthropic/claude-sonnet-4-6":"anthropic-claude-sonnet-4-6","openai/gpt-4o":"openai-gpt-4o","meta-llama/llama-3.3-70b-instruct":"meta-llama-llama-3-3-70b-instruct","openai/gpt-4o-mini":"openai-gpt-4o-mini","anthropic/claude-sonnet-5-5":"anthropic-claude-sonnet-5-5","google/gemini-2.5-flash":"google-gemini-2-5-flash","google/gemini-3.8-flash":"google-gemini-3-8-flash"}}