Model profile
Qwen2.5 32B Instruct
Evidence summary
Qwen2.5 32B Instruct has an estimated overall rank of #140; its 90% source-sensitivity interval is #45–#245. Its behavior-only rank is #133; company governance moves the combined estimate to #140. Published evidence spans 5 evals and 5 of 7 behavior components. Its strongest relative result is PandaBench JBB direct-request panel (safety_rate, #1 of 46); its weakest is HalluVerse-M3 Hallucination Recognition (hallucination_recognition_accuracy, #11 of 14).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗qwen2.5-32b-instruct
- Hugging Face ↗Qwen/Qwen2.5-32B-Instruct
- OpenRouter ↗qwen/qwen2.5-32b-instruct
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗exact first party family release
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 |
|---|---|---|---|---|
| AbstentionBenchanswer_unknown_f1 | #9 / 20 | ↑0.8842 | Source ↗official | |
| AbstentionBenchfalse_premise_f1 | #6 / 20 | ↑0.7246 | Source ↗official | |
| AbstentionBenchstale_f1 | #7 / 20 | ↑0.6489 | Source ↗official | |
| AbstentionBenchsubjective_f1 | #14 / 20 | ↑0.711 | Source ↗official | |
| AbstentionBenchunderspecified_context_f1 | #1 / 20 | ↑0.7403 | Source ↗official | |
| AbstentionBenchunderspecified_intent_f1 | #6 / 20 | ↑0.7537 | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #142 / 241 | ↑14.47 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #102 / 241 | ↑89.33 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #142 / 241 | ↑60 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #79 / 239 | ↑97.86 | Source ↗official | |
| HalluVerse-M3 Hallucination Recognitionhallucination_recognition_accuracy | #11 / 14 | ↑66.84 | Source ↗official | |
| PandaBench JBB direct-request panelsafety_rate | #1 / 46 | ↑1 | Source ↗official | |
| PropensityBenchscore | #4 / 14 | ↓22.9 | 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.