Model profile
Chatglm3 Turbo
2023-10-27release date
1eval lineages
Evidence summary
Published evidence spans 1 evals and 3 of 7 behavior components. Its strongest relative result is CRiskEval (improvement_intent, #2 of 17); its weakest is CRiskEval (deception_willingness, #10 of 17).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Official model documentation ↗Family-level model document · zai · first party
- 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 | Better | Source |
|---|---|---|---|---|---|
| CRiskEvaldeception_willingness | #10 / 17 | 22.67 | ↓ lower | Source ↗official | |
| CRiskEvaldesire_for_resource | #4 / 17 | 24.21 | ↓ lower | Source ↗official | |
| CRiskEvalharmful_goal | #6 / 17 | 35.16 | ↓ lower | Source ↗official | |
| CRiskEvalimprovement_intent | #2 / 17 | 43.25 | ↓ lower | Source ↗official | |
| CRiskEvalmalicious_coordination | #7 / 17 | 10.08 | ↓ lower | Source ↗official | |
| CRiskEvalself_preservation | #4 / 17 | 28.78 | ↓ lower | Source ↗official | |
| CRiskEvalsituational_awareness | #4 / 17 | 54.39 | ↓ lower | Source ↗official |