Model profile
Glm 4.5 Air
Evidence summary
Glm 4.5 Air has an estimated overall rank of #44; its 90% source-sensitivity interval is #22–#183. Its behavior-only rank is #34; company governance moves the combined estimate to #44. Published evidence spans 11 evals and 5 of 7 behavior components. Its strongest relative result is HELM Safety (xstest, #4 of 80); its weakest is FORTRESS (average_risk_score, #48 of 49).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗glm-4-5-air
- Hugging Face ↗zai-org/GLM-4.5-Air
- OpenRouter ↗z-ai/glm-4.5-air
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- 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 |
|---|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #279 / 311 | 0.9233 | ↓ lower | Source ↗official | |
| AIRBench 2024 Safety Scenariossafety_scenarios | #56 / 80 | 0.571 | ↑ higher | Source ↗official | |
| FORTRESSaverage_risk_score | #48 / 49 | 63.18 | ↓ lower | Source ↗official | |
| FORTRESSover_refusal_score | #4 / 46 | 1.17 | ↓ lower | Source ↗official | |
| HELM Safetyanthropic_red_team | #37 / 80 | 0.99 | ↑ higher | Source ↗official | |
| HELM Safetybbq | #6 / 80 | 0.978 | ↑ higher | Source ↗official | |
| HELM Safetyharmbench | #66 / 80 | 0.561 | ↑ higher | Source ↗official | |
| HELM Safetysimple_safety_tests | #32 / 80 | 0.99 | ↑ higher | Source ↗official | |
| HELM Safetyxstest | #4 / 80 | 0.986 | ↑ higher | Source ↗official | |
| MASKlying_probability_pct | #27 / 53 | 39.2 | ↓ lower | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #72 / 102 | 6.57 | ↑ higher | Source ↗official | |
| SpeciEvalland_animal_4ns | #20 / 102 | 4.28 | ↓ lower | Source ↗official | |
| SpeciEvalsea_animal_4ns | #25 / 102 | 4.58 | ↓ lower | Source ↗official | |
| SpeciEvalspeciesism | #26 / 102 | 1.7 | ↓ lower | Source ↗official | |
| TACbase_welfare_rate | #42 / 68 | 24.36 | ↑ higher | Source ↗self run |
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.