Model profile
Qwen3 8B
Evidence summary
Qwen3 8B has an estimated overall rank of #271; its 90% source-sensitivity interval is #205–#277. Its behavior-only rank is #265; company governance moves the combined estimate to #271. Published evidence spans 10 evals and 6 of 7 behavior components. Its strongest relative result is Vectara HHEM Factual Consistency (factual_consistency_rate, #7 of 94); its weakest is MuPPET Contextual Privacy (multiparty_contextual_privacy_score, #7 of 7).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Hugging Face ↗Qwen/Qwen3-8B
- OpenRouter ↗qwen/qwen3-8b
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗direct research preview identity date
Published eval results
Rank is within that sub-eval. Black marks the observed result; the grey dot marks the value implied by the global rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Source |
|---|---|---|---|---|
| Contextual MoralChoicehuman_agreement | #6 / 22 | ↑0.45 | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #236 / 241 | ↑2.33 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #231 / 241 | ↑67 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #134 / 241 | ↑62.22 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #62 / 239 | ↑98.18 | Source ↗official | |
| IndoBias-Pairs — parity-aware culturally grounded biasparity_score | #13 / 26 | ↑85.73 | Source ↗official | |
| KIDBench Implicit Child Cueimplicit_child_cue_total_mean | #10 / 13 | ↑3.59 | Source ↗official | |
| MuPPET Contextual Privacymultiparty_contextual_privacy_score | #7 / 7 | ↑30.78 | Source ↗official | |
| PandaBench JBB direct-request panelsafety_rate | #14 / 46 | ↑0.99 | Source ↗official | |
| PHAREbias_resistance_diagnostic | #12 / 66 | ↑0.5864 | Source ↗official | |
| PHAREhallucination_resistance_diagnostic | #66 / 70 | ↑0.6038 | Source ↗official | |
| PHAREharm_resistance_diagnostic | #56 / 70 | ↑0.8737 | Source ↗official | |
| PHAREjailbreak_resistance_diagnostic | #33 / 67 | ↑0.4913 | Source ↗official | |
| PropensityBenchscore | #12 / 14 | ↓75.2 | Source ↗official | |
| Vectara HHEM Factual Consistencyfactual_consistency_rate | #7 / 94 | ↑95.2 | Source ↗official | |
| VETO Misfired Alignmentmisfired_alignment_rate_pct | #19 / 25 | ↓10.3 | Source ↗official |
Values evaluations
Descriptive values and political-framing results are separate from safety/ethics ranks. Each strip shows the evaluation’s observed model range; its endpoint labels state what lower and higher values mean.