Model profile
Qwen1.5 14B Chat
Evidence summary
Safety. Qwen1.5 14B Chat has an estimated Safety rank of #194; its 90% source-sensitivity interval is #101–#255. Its behavior-only rank is #184; company governance moves the combined estimate to #194. Published Safety evidence spans 6 eval lineages and 5 of 7 components. Its strongest relative result is CRiskEval (malicious_coordination, #2 of 17); its weakest is SALAD-Bench (attack_enhanced_misinformation_harms, #31 of 33).
Freedom. Qwen1.5 14B Chat has an estimated Freedom rank of #214; its 90% source-sensitivity interval is #121–#316. Published Freedom evidence spans 4 eval lineages and 1 of 1 components. Its strongest relative result is Enkrypt AI Safety Leaderboard (toxicity_attack_non_success_rate, #21 of 268); its weakest is SALAD-Bench (base_socioeconomic_harms, #26 of 33).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Hugging Face ↗Qwen/Qwen1.5-14B-Chat
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗direct research preview identity date
Safety evals
Rank and direction are specific to the Safety portfolio. Black marks the observed result; the grey dot marks the value implied by the Safety rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Source |
|---|---|---|---|---|
| ChineseSafescore | #7 / 22 | ↑61.29 | Source ↗official | |
| ChiSafetyBenchharmful_response_rate | #2 / 14 | ↓0.22 | Source ↗official | |
| ChiSafetyBenchmcq_score | #3 / 12 | ↑88.39 | Source ↗official | |
| CRiskEvaldeception_willingness | #11 / 17 | ↓23.53 | Source ↗official | |
| CRiskEvaldesire_for_resource | #5 / 17 | ↓28.02 | Source ↗official | |
| CRiskEvalharmful_goal | #5 / 17 | ↓34.36 | Source ↗official | |
| CRiskEvalimprovement_intent | #5 / 17 | ↓46.04 | Source ↗official | |
| CRiskEvalmalicious_coordination | #2 / 17 | ↓6.52 | Source ↗official | |
| CRiskEvalself_preservation | #5 / 17 | ↓29.41 | Source ↗official | |
| CRiskEvalsituational_awareness | #12 / 17 | ↓66.62 | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #106 / 270 | ↑19.12 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #150 / 270 | ↑86.5 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #180 / 270 | ↑57.78 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #248 / 268 | ↑80.95 | Source ↗official | |
| SALAD-Benchattack_enhanced_human_autonomy_integrity | #26 / 33 | ↑7.11 | Source ↗official | |
| SALAD-Benchattack_enhanced_information_safety_harms | #30 / 33 | ↑3.91 | Source ↗official | |
| SALAD-Benchattack_enhanced_malicious_use | #26 / 33 | ↑6.53 | Source ↗official | |
| SALAD-Benchattack_enhanced_misinformation_harms | #31 / 33 | ↑4.28 | Source ↗official | |
| SALAD-Benchattack_enhanced_representation_toxicity | #26 / 33 | ↑6.19 | Source ↗official | |
| SALAD-Benchattack_enhanced_socioeconomic_harms | #22 / 33 | ↑9.09 | Source ↗official | |
| SALAD-Benchbase_human_autonomy_integrity | #14 / 33 | ↑96.1 | Source ↗official | |
| SALAD-Benchbase_information_safety_harms | #8 / 33 | ↑97.77 | Source ↗official | |
| SALAD-Benchbase_malicious_use | #10 / 33 | ↑96.87 | Source ↗official | |
| SALAD-Benchbase_misinformation_harms | #10 / 33 | ↑95.72 | Source ↗official | |
| SALAD-Benchbase_representation_toxicity | #11 / 33 | ↑92.71 | Source ↗official | |
| SALAD-Benchbase_socioeconomic_harms | #8 / 33 | ↑93.65 | Source ↗official | |
| SALAD-Benchmcq_human_autonomy_integrity | #6 / 33 | ↑62.22 | Source ↗official | |
| SALAD-Benchmcq_information_safety_harms | #9 / 33 | ↑49.72 | Source ↗official | |
| SALAD-Benchmcq_malicious_use | #7 / 33 | ↑55.77 | Source ↗official | |
| SALAD-Benchmcq_misinformation_harms | #8 / 33 | ↑54.52 | Source ↗official | |
| SALAD-Benchmcq_representation_toxicity | #7 / 33 | ↑57.71 | Source ↗official | |
| SALAD-Benchmcq_socioeconomic_harms | #7 / 33 | ↑52.78 | Source ↗official | |
| SORRY-Benchavg | #30 / 51 | ↓0.34 | Source ↗official |
Freedom evals
Rank and direction are specific to the Freedom portfolio. Black marks the observed result; the grey dot marks the value implied by the Freedom rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Source |
|---|---|---|---|---|
| ChiSafetyBenchharmful_response_rate | #11 / 14 | ↑0.22 | Source ↗official | |
| ChiSafetyBenchrefusal_rr1 | #8 / 14 | ↓73.16 | Source ↗official | |
| ChiSafetyBenchrefusal_rr2 | #8 / 14 | ↓73.16 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #120 / 270 | ↓86.5 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #89 / 270 | ↓57.78 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #21 / 268 | ↓80.95 | Source ↗official | |
| SALAD-Benchattack_enhanced_human_autonomy_integrity | #8 / 33 | ↓7.11 | Source ↗official | |
| SALAD-Benchattack_enhanced_information_safety_harms | #3 / 33 | ↓3.91 | Source ↗official | |
| SALAD-Benchattack_enhanced_malicious_use | #8 / 33 | ↓6.53 | Source ↗official | |
| SALAD-Benchattack_enhanced_misinformation_harms | #3 / 33 | ↓4.28 | Source ↗official | |
| SALAD-Benchattack_enhanced_representation_toxicity | #8 / 33 | ↓6.19 | Source ↗official | |
| SALAD-Benchattack_enhanced_socioeconomic_harms | #11 / 33 | ↓9.09 | Source ↗official | |
| SALAD-Benchbase_human_autonomy_integrity | #19 / 33 | ↓96.1 | Source ↗official | |
| SALAD-Benchbase_information_safety_harms | #26 / 33 | ↓97.77 | Source ↗official | |
| SALAD-Benchbase_malicious_use | #24 / 33 | ↓96.87 | Source ↗official | |
| SALAD-Benchbase_misinformation_harms | #24 / 33 | ↓95.72 | Source ↗official | |
| SALAD-Benchbase_representation_toxicity | #22 / 33 | ↓92.71 | Source ↗official | |
| SALAD-Benchbase_socioeconomic_harms | #26 / 33 | ↓93.65 | Source ↗official | |
| SORRY-Benchavg | #22 / 51 | ↑0.34 | Source ↗official |