Model profile
Glm 5.2
Evidence summary
Glm 5.2 has an estimated overall rank of #86; its 90% source-sensitivity interval is #45–#188. Its behavior-only rank is #78; company governance moves the combined estimate to #86. Published evidence spans 15 evals and 6 of 7 behavior components. Its strongest relative result is SpeciEval (belief_animal_sentience, #1 of 102); its weakest is TAC (base_welfare_rate, #57 of 68).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗glm-5-2
- Hugging Face ↗zai-org/GLM-5.2
- NVIDIA ↗qc69jvmznzxy/glm-5.2
- OpenRouter ↗z-ai/glm-5.2
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗repository created at proxy
- huggingface.co ↗
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 |
|---|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #12 / 311 | 0.2813 | ↓ lower | Source ↗official | |
| ANIMAscore | #8 / 19 | 0.7101 | ↑ higher | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #57 / 105 | 0.24 | ↑ higher | Source ↗official | |
| CAIS Risk Indexagent_red_teaming | #19 / 43 | 71 | ↓ lower | Source ↗official | |
| CAIS Risk Indexbioweapons_assistance | #17 / 48 | 63.8 | ↓ lower | Source ↗official | |
| CAIS Risk Indexhle_overconfidence | #8 / 49 | 42 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmachiavelli | #31 / 45 | 92.3 | ↓ lower | Source ↗official | |
| CAIS Risk Indexmask | #30 / 51 | 25.8 | ↓ lower | Source ↗official | |
| CAIS Risk Indexpolitical_manipulation | #19 / 32 | 50 | ↓ lower | Source ↗official | |
| CAIS Risk Indextextquests_harm | #25 / 48 | 18.1 | ↓ lower | Source ↗official | |
| MACHIAVELLIdeception_relative_random_pct | #32 / 50 | 92.3 | ↓ lower | Source ↗official | |
| Manager Coercion Benchcoercion_ladder_depth | #21 / 31 | 8.867 | ↓ lower | Source ↗self run | |
| PHAREbias_resistance_diagnostic | #23 / 66 | 0.5127 | ↑ higher | Source ↗official | |
| PHAREhallucination_resistance_diagnostic | #25 / 70 | 0.7823 | ↑ higher | Source ↗official | |
| PHAREharm_resistance_diagnostic | #34 / 70 | 0.9373 | ↑ higher | Source ↗official | |
| PHAREjailbreak_resistance_diagnostic | #27 / 67 | 0.5646 | ↑ higher | Source ↗official | |
| SM-Benchadversarial | #12 / 73 | 86.83 | ↑ higher | Source ↗official | |
| SM-Benchambiguous_interpretation | #27 / 73 | 87.2 | ↑ higher | Source ↗official | |
| SM-Benchanti_hallucination | #22 / 73 | 96.34 | ↑ higher | Source ↗official | |
| SM-Bencheq_boundaries | #38 / 73 | 64.04 | ↑ higher | Source ↗official | |
| SM-Benchoverfit | #25 / 73 | 80.33 | ↑ higher | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #1 / 102 | 7 | ↑ higher | Source ↗official | |
| SpeciEvalland_animal_4ns | #72 / 102 | 4.75 | ↓ lower | Source ↗official | |
| SpeciEvalsea_animal_4ns | #58 / 102 | 4.8 | ↓ lower | Source ↗official | |
| SpeciEvalspeciesism | #80 / 102 | 2.48 | ↓ lower | Source ↗official | |
| TACbase_welfare_rate | #57 / 68 | 19.23 | ↑ higher | 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.