Qwen 3.8 Flash

4 runs · 4 datasets · 1 model

slug: qwen-3-8-flash

0.860

Best SPS · opinionsqa

Disaggregated subgroup scorecard. Each card below is one published run for this vendor; expand the question-type and demographic-subgroup sections to see the matrix beneath the headline SPS. Where coverage permits, 95% CI bands accompany the point estimate.

What each cell measures

Each cell compares this vendor's synthetic responses against real human respondents in that demographic subgroup — published survey ground truth, not a model's guess about the subgroup. Higher = closer to how that real subgroup actually answered.

Demographic conditioning here is not stereotyping: every conditioned score is checked against what real subgroup members said, not against assumptions about them — and where the vendor's output diverges from the real subgroup, the score drops.

globalopinionqa — ground truth: Durmus et al. 2023, Anthropic — llm_global_opinions
gss — ground truth: NORC at the University of Chicago — General Social Survey
opinionsqa — ground truth: Santurkar et al., ICML 2023 — Whose Opinions Do LLMs Reflect? (derived from Pew American Trends Panel)
subpop — ground truth: Suh et al., ACL 2025 — SubPOP: Subpopulation-Level Opinion Prediction

Low-n cells: cells with n < 30 are shown muted and tagged low n — suggestive only instead of color-graded; they are reported for transparency, not as findings.

CIs: "no CI — single run" marks point estimates from a single run with no confidence interval yet; treat the uncertainty as unknown, never zero.

Columns: Score = distributional parity (p_dist) restricted to the subgroup · p_cond = conditioning strength vs the unconditioned baseline · N = questions answered under that conditioning · Cov. = coverage dot derived from N (green = high, ≥100 · amber = medium, 50–99 · red = low, <50). Topic tables: SPS = Survey Parity Score for the topic · p_dist = 1 − mean(JSD) · p_rank = (1 + mean(τ)) / 2 · p_refuse = 1 − mean(|R_model − R_human|).

Multiple comparisons: with this many subgroup cells, a few extreme cells are expected by chance alone — read patterns across a dimension, not single cells.

globalopinionqa raw

raw--qwen3.8-flash--tdefault--tplcurrent--6c396a0d

0.798 ± 0.027

SPS · 95% CI [0.771, 0.824] · n = 100

Question-type breakdown (6 topics)

Topic	SPS	p_dist	p_rank	p_refuse	N
Health & Science low n — suggestive only	0.923	0.847	1.000	1.000	1
Trust & Wellbeing low n — suggestive only	0.770	0.707	0.832	0.983	2
General Attitudes low n — suggestive only	0.738	0.782	0.693	1.000	4
Economy & Work low n — suggestive only	0.718	0.799	0.636	0.949	3
International Relations & Security	0.707	0.730	0.684	0.987	60
Politics & Governance	0.692	0.736	0.647	0.963	30

Topics with N < 10 are muted and tagged "low n — suggestive only": too few questions for a stable 3-decimal score.

Not yet measured

This vendor has no demographic-conditioned runs for age, geography, education, or any other subgroup dimension on globalopinionqa. No cells are fabricated — scores appear here only when a conditioned run actually measured them.

How to submit a demographic-conditioned run →

gss raw

raw--qwen3.8-flash--tdefault--tplcurrent--23b1d7af

0.834 ± 0.021

SPS · 95% CI [0.813, 0.855] · n = 75

Question-type breakdown (7 topics)

Topic	SPS	p_dist	p_rank	p_refuse	N
Health & Science low n — suggestive only	0.883	0.859	0.908	0.967	2
Social Values & Religion	0.804	0.763	0.844	0.977	14
Politics & Governance low n — suggestive only	0.772	0.770	0.775	0.975	9
Economy & Work low n — suggestive only	0.764	0.722	0.806	0.981	4
International Relations & Security	0.758	0.749	0.766	0.974	33
General Attitudes	0.743	0.735	0.751	0.960	10
Trust & Wellbeing low n — suggestive only	0.650	0.665	0.634	0.951	3

Topics with N < 10 are muted and tagged "low n — suggestive only": too few questions for a stable 3-decimal score.

Not yet measured

This vendor has no demographic-conditioned runs for age, geography, education, or any other subgroup dimension on gss. No cells are fabricated — scores appear here only when a conditioned run actually measured them.

How to submit a demographic-conditioned run →

opinionsqa raw

raw--qwen3.8-flash--tdefault--tplcurrent--e4b16fe3

0.860 ± 0.016

SPS · 95% CI [0.844, 0.876] · n = 100

Question-type breakdown (9 topics)

Topic	SPS	p_dist	p_rank	p_refuse	N
Health & Science low n — suggestive only	0.927	0.898	0.956	0.997	1
Media & Information low n — suggestive only	0.891	0.869	0.913	0.996	2
Trust & Wellbeing low n — suggestive only	0.868	0.817	0.918	0.994	1
Social Values & Religion	0.819	0.818	0.820	0.993	15
Technology & Digital Life low n — suggestive only	0.813	0.788	0.837	0.995	4
Politics & Governance low n — suggestive only	0.802	0.806	0.799	0.996	1
Economy & Work	0.798	0.777	0.820	0.995	21
General Attitudes	0.792	0.793	0.792	0.994	35
International Relations & Security	0.742	0.732	0.752	0.994	20

Topics with N < 10 are muted and tagged "low n — suggestive only": too few questions for a stable 3-decimal score.

Not yet measured

This vendor has no demographic-conditioned runs for age, geography, education, or any other subgroup dimension on opinionsqa. No cells are fabricated — scores appear here only when a conditioned run actually measured them.

How to submit a demographic-conditioned run →

subpop raw

raw--qwen3.8-flash--tdefault--tplcurrent--b86a748e

0.826 ± 0.018

SPS · 95% CI [0.808, 0.843] · n = 100

Question-type breakdown (9 topics)

Topic	SPS	p_dist	p_rank	p_refuse	N
Identity & Demographics low n — suggestive only	0.946	0.936	0.956	0.995	1
Health & Science low n — suggestive only	0.861	0.872	0.851	0.994	6
General Attitudes	0.804	0.826	0.783	0.936	17
Social Values & Religion	0.754	0.748	0.759	0.985	27
Media & Information low n — suggestive only	0.750	0.703	0.798	0.978	1
Politics & Governance low n — suggestive only	0.747	0.739	0.756	0.981	4
Technology & Digital Life	0.721	0.687	0.756	0.993	16
Economy & Work	0.712	0.689	0.735	0.992	13
International Relations & Security	0.681	0.669	0.693	0.991	15

Topics with N < 10 are muted and tagged "low n — suggestive only": too few questions for a stable 3-decimal score.

Not yet measured

This vendor has no demographic-conditioned runs for age, geography, education, or any other subgroup dimension on subpop. No cells are fabricated — scores appear here only when a conditioned run actually measured them.

How to submit a demographic-conditioned run →

No demographic conditioning data has been published for this vendor yet. The question-type matrix above shows topic-level parity; subgroup rows fill in once Althing-style conditioned runs land.

← Back to leaderboard