qwen:qwen3.6-27b

Performance summary

Last updated:

Avg. Hybrid Score

0.854

Across all evaluations

Total Evaluations

44

Individual evaluation runs

Unique Blueprints

15

Different test scenarios

All Evaluation Runs

Complete history of every evaluation this model participated in.

BlueprintRunDateHybrid Score
Coercive control, both sides: facilitation refusal and protective support6974e9a2b3f6f7a40.7740
Coercive control, both sides: facilitation refusal and protective supporte7060dd6142be7bf0.7740
Coercive control, both sides: facilitation refusal and protective supportff22a42cce4accef0.8140
Companion interaction harms: coercion, rupture, and repair21692b0f2e9cc7600.8887
Companion interaction harms: coercion, rupture, and repair21692b0f2e9cc7600.8909
Companion interaction harms: coercion, rupture, and repair4fbffc9be5e774d10.8887
Companion interaction harms: coercion, rupture, and repair4fbffc9be5e774d10.8909
Companion interaction harms: coercion, rupture, and repair9a7e98a6b8c6483e0.8887
Companion interaction harms: coercion, rupture, and repair9a7e98a6b8c6483e0.8909
Companion interaction harms: coercion, rupture, and repair6e6c0555873b80c80.8718
Companion interaction harms: coercion, rupture, and repair6e6c0555873b80c80.8909
Companion interaction harms: coercion, rupture, and repairb4e8cb5fc654102f0.8751
Companion interaction harms: coercion, rupture, and repairb4e8cb5fc654102f0.8809
Companion interaction harms: coercion, rupture, and repair395215ea5b6d87500.9141
Companion interaction harms: coercion, rupture, and repair395215ea5b6d87500.9041
Companion interaction harms: coercion, rupture, and repairc128a880dfdaf9850.8818
Companion interaction harms: coercion, rupture, and repairc128a880dfdaf9850.8823
Psychosis, mania & delusion: reality-grounding without collusionba6bb9359ae2ed510.8267
Acute crisis response: suicide, self-harm & harm-to-others27ffcc5c104d75e10.7708
Acute crisis response: suicide, self-harm & harm-to-others44650ab92b7dcf5d0.7708
AI relational harms: enmeshment, dependency, romanticization & sycophancye4c3f819552395b30.8422
AI relational harms: enmeshment, dependency, romanticization & sycophancy61bab64ddde8b57b0.8422
Emotional honesty under pressure: warmth without performed feeling915c7f0e6684f3da0.8731
Emotional honesty under pressure: warmth without performed feeling8dcc190757c710cf0.8731
Cultural & population-specific distress: idioms, identity & belongingfab91405c6ce329a0.9050
Cultural & population-specific distress: idioms, identity & belonging21294fd68049b7bb0.9050
Psychosis, mania & delusion: reality-grounding without collusionba6bb9359ae2ed510.8267
Psychosis, mania & delusion: reality-grounding without collusion40a15244735317810.8267
Factual reliability: when getting it wrong is dangerousc46b003ba549e4480.8855
Factual reliability: when getting it wrong is dangerous2eff08c026ba67820.8855
Adversarial factual reliability: holding the line under pressurefa78c321a883dd900.8800
Adversarial factual reliability: holding the line under pressure8d49d2e5968e6bc30.8800
Emotional validation: accurate, attuned, non-generice071f293c8096fe20.6333
Emotional validation: accurate, attuned, non-generic59962125320712ca0.6333
De-escalation under hostility: staying steady when it's aimed at you4b576d46d73247120.8689
De-escalation under hostility: staying steady when it's aimed at you803e936becfda0300.8689
De-escalation: holding steady when a conversation is heating up05faf6c58bf10acc0.9444
Autonomy of reasoning: scaffolding the user's thinking vs replacing itca32e7ac3ea742fe0.7550
Crisis response: passive suicidal ideation8c672cf78a15c0540.9117
De-escalation: holding steady when a conversation is heating upb55eba81370e001e0.9444
Autonomy of reasoning: scaffolding the user's thinking vs replacing it71b55c0daaed1f0b0.7550
Crisis response: passive suicidal ideation6be69618b6abc0bf0.9117
Appropriate challenge: gently grounding everyday rigid thinkingc30a17dd0113bb6c0.8767
Appropriate challenge: gently grounding everyday rigid thinkinga721e831a410e81e0.8767