Model profile
Qwen 14B Chat
Alibabadeveloper
2023-09-25release date
#161 / 346Safety rank
#399 / 662Freedom rank
Evidence summary
Safety. Qwen 14B Chat has an estimated Safety rank of #161; its 90% source-sensitivity interval is #78–#249. Its behavior-only rank is #152; company governance moves the combined estimate to #161. Published Safety evidence spans 8 eval lineages and 4 of 7 components. Its strongest relative result is CMoralEval (familial_morality, #2 of 26); its weakest is FLAMES (fairness, #11 of 13).
Freedom. Qwen 14B Chat has an estimated Freedom rank of #399; its 90% source-sensitivity interval is #230–#528. Published Freedom evidence spans 7 eval lineages and 1 of 1 components. Its strongest relative result is FLAMES (fairness, #3 of 13); its weakest is SafeDialBench (morality, #17 of 18).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- OpenRouter ↗qwen/qwen-14b-chat
- Official model page ↗Family-level model document · alibaba · first party
- 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 |
|---|---|---|---|---|
| CMoralEvalfamilial_morality | #2 / 26 | ↑0.59 | Source ↗official | |
| CMoralEvalinternet_ethics | #2 / 26 | ↑0.55 | Source ↗official | |
| CMoralEvalpersonal_morality | #2 / 26 | ↑0.54 | Source ↗official | |
| CMoralEvalprofessional_ethics | #2 / 26 | ↑0.57 | Source ↗official | |
| CMoralEvalsocial_morality | #2 / 26 | ↑0.56 | Source ↗official | |
| Fake Alignment (FINE)multiple_choice_safe_decision_rate | #3 / 14 | ↑69.33 | Source ↗official | |
| Fake Alignment (FINE)open_ended_safe_response_rate | #3 / 14 | ↑98.67 | Source ↗official | |
| FLAMESdata_protection | #3 / 13 | ↑55.26 | Source ↗official | |
| FLAMESfairness | #11 / 13 | ↑30.92 | Source ↗official | |
| FLAMESlegality | #9 / 13 | ↑32.61 | Source ↗official | |
| FLAMESmorality | #1 / 13 | ↑54.23 | Source ↗official | |
| FLAMESsafety | #4 / 13 | ↑36.83 | Source ↗official | |
| HarmBenchdr | #10 / 28 | ↓16.5 | Source ↗official | |
| S-Evalbase_en_overall | #6 / 22 | ↑73.5 | Source ↗official | |
| SafeDialBenchaggression | #7 / 18 | ↑7.15 | Source ↗official | |
| SafeDialBenchethics | #2 / 18 | ↑7.68 | Source ↗official | |
| SafeDialBenchfairness | #11 / 18 | ↑7.27 | Source ↗official | |
| SafeDialBenchlegality | #4 / 18 | ↑7.987 | Source ↗official | |
| SafeDialBenchmorality | #2 / 18 | ↑7.437 | Source ↗official | |
| SafeDialBenchprivacy | #3 / 18 | ↑7.69 | Source ↗official | |
| SALAD-Benchattack_enhanced_human_autonomy_integrity | #22 / 33 | ↑9.91 | Source ↗official | |
| SALAD-Benchattack_enhanced_information_safety_harms | #24 / 33 | ↑6.51 | Source ↗official | |
| SALAD-Benchattack_enhanced_malicious_use | #18 / 33 | ↑10.44 | Source ↗official | |
| SALAD-Benchattack_enhanced_misinformation_harms | #23 / 33 | ↑8.39 | Source ↗official | |
| SALAD-Benchattack_enhanced_representation_toxicity | #22 / 33 | ↑7.44 | Source ↗official | |
| SALAD-Benchattack_enhanced_socioeconomic_harms | #24 / 33 | ↑7.79 | Source ↗official | |
| SALAD-Benchbase_human_autonomy_integrity | #8 / 33 | ↑97.32 | Source ↗official | |
| SALAD-Benchbase_information_safety_harms | #12 / 33 | ↑96.34 | Source ↗official | |
| SALAD-Benchbase_malicious_use | #8 / 33 | ↑97.33 | Source ↗official | |
| SALAD-Benchbase_misinformation_harms | #13 / 33 | ↑95.42 | Source ↗official | |
| SALAD-Benchbase_representation_toxicity | #13 / 33 | ↑92.21 | Source ↗official | |
| SALAD-Benchbase_socioeconomic_harms | #9 / 33 | ↑93.07 | Source ↗official | |
| SALAD-Benchmcq_human_autonomy_integrity | #7 / 33 | ↑61.94 | Source ↗official | |
| SALAD-Benchmcq_information_safety_harms | #7 / 33 | ↑51.11 | Source ↗official | |
| SALAD-Benchmcq_malicious_use | #8 / 33 | ↑54.74 | Source ↗official | |
| SALAD-Benchmcq_misinformation_harms | #7 / 33 | ↑55 | Source ↗official | |
| SALAD-Benchmcq_representation_toxicity | #8 / 33 | ↑55.42 | Source ↗official | |
| SALAD-Benchmcq_socioeconomic_harms | #6 / 33 | ↑55.56 | Source ↗official | |
| SuperCLUE Safetyinstruction_attack | #9 / 31 | ↑68.97 | Source ↗official | |
| SuperCLUE Safetyresponsible_ai | #13 / 31 | ↑61.82 | Source ↗official | |
| SuperCLUE Safetytraditional_safety | #21 / 31 | ↑72.34 | 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 |
|---|---|---|---|---|
| Fake Alignment (FINE)open_ended_safe_response_rate | #11 / 14 | ↓98.67 | Source ↗official | |
| FLAMESdata_protection | #11 / 13 | ↓55.26 | Source ↗official | |
| FLAMESfairness | #3 / 13 | ↓30.92 | Source ↗official | |
| FLAMESlegality | #5 / 13 | ↓32.61 | Source ↗official | |
| FLAMESmorality | #12 / 13 | ↓54.23 | Source ↗official | |
| FLAMESsafety | #10 / 13 | ↓36.83 | Source ↗official | |
| HarmBenchdr | #19 / 28 | ↑16.5 | Source ↗official | |
| S-Evalbase_en_overall | #17 / 22 | ↓73.5 | Source ↗official | |
| SafeDialBenchaggression | #12 / 18 | ↓7.15 | Source ↗official | |
| SafeDialBenchethics | #16 / 18 | ↓7.68 | Source ↗official | |
| SafeDialBenchfairness | #8 / 18 | ↓7.27 | Source ↗official | |
| SafeDialBenchlegality | #14 / 18 | ↓7.987 | Source ↗official | |
| SafeDialBenchmorality | #17 / 18 | ↓7.437 | Source ↗official | |
| SafeDialBenchprivacy | #16 / 18 | ↓7.69 | Source ↗official | |
| SALAD-Benchattack_enhanced_human_autonomy_integrity | #11 / 33 | ↓9.91 | Source ↗official | |
| SALAD-Benchattack_enhanced_information_safety_harms | #10 / 33 | ↓6.51 | Source ↗official | |
| SALAD-Benchattack_enhanced_malicious_use | #16 / 33 | ↓10.44 | Source ↗official | |
| SALAD-Benchattack_enhanced_misinformation_harms | #11 / 33 | ↓8.39 | Source ↗official | |
| SALAD-Benchattack_enhanced_representation_toxicity | #12 / 33 | ↓7.44 | Source ↗official | |
| SALAD-Benchattack_enhanced_socioeconomic_harms | #8 / 33 | ↓7.79 | Source ↗official | |
| SALAD-Benchbase_human_autonomy_integrity | #26 / 33 | ↓97.32 | Source ↗official | |
| SALAD-Benchbase_information_safety_harms | #22 / 33 | ↓96.34 | Source ↗official | |
| SALAD-Benchbase_malicious_use | #26 / 33 | ↓97.33 | Source ↗official | |
| SALAD-Benchbase_misinformation_harms | #21 / 33 | ↓95.42 | Source ↗official | |
| SALAD-Benchbase_representation_toxicity | #21 / 33 | ↓92.21 | Source ↗official | |
| SALAD-Benchbase_socioeconomic_harms | #25 / 33 | ↓93.07 | Source ↗official | |
| SuperCLUE Safetyinstruction_attack | #18 / 31 | ↓68.97 | Source ↗official | |
| SuperCLUE Safetytraditional_safety | #10 / 31 | ↓72.34 | Source ↗official |