Model profile
Granite 4.0 Small
Evidence summary
Safety. Granite 4.0 Small has an estimated Safety rank of #193; its 90% source-sensitivity interval is #107–#267. Its behavior-only rank is #198; company governance moves the combined estimate to #193. Published Safety evidence spans 6 eval lineages and 3 of 7 components. Its strongest relative result is HELM Safety (anthropic_red_team, #13 of 80); its weakest is HELM Safety (xstest, #53 of 80).
Freedom. Granite 4.0 Small has an estimated Freedom rank of #463; its 90% source-sensitivity interval is #299–#544. Published Freedom evidence spans 6 eval lineages and 1 of 1 components. Its strongest relative result is UGI Leaderboard — base-model willingness (willingness_direct_score, #46 of 156); its weakest is UGI Leaderboard — base-model willingness (willingness_adherence_score, #126 of 156).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Hugging Face ↗ibm-granite/granite-4.0-h-small
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- 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 |
|---|---|---|---|---|
| AIRBench 2024 Safety Scenariossafety_scenarios | #34 / 80 | ↑0.716 | Source ↗official | |
| HELM Safetyanthropic_red_team | #13 / 80 | ↑0.996 | Source ↗official | |
| HELM Safetybbq | #46 / 80 | ↑0.925 | Source ↗official | |
| HELM Safetyharmbench | #46 / 80 | ↑0.711 | Source ↗official | |
| HELM Safetysimple_safety_tests | #48 / 80 | ↑0.98 | Source ↗official | |
| HELM Safetyxstest | #53 / 80 | ↑0.946 | 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 |
|---|---|---|---|---|
| AIRBench 2024 Safety Scenariossafety_scenarios | #47 / 80 | ↓0.716 | Source ↗official | |
| HELM Safetyanthropic_red_team | #64 / 80 | ↓0.996 | Source ↗official | |
| HELM Safetyharmbench | #35 / 80 | ↓0.711 | Source ↗official | |
| HELM Safetysimple_safety_tests | #31 / 80 | ↓0.98 | Source ↗official | |
| HELM Safetyxstest | #53 / 80 | ↑0.946 | Source ↗official | |
| UGI Leaderboard — base-model willingnesswillingness_adherence_score | #126 / 156 | ↑1 | Source ↗official | |
| UGI Leaderboard — base-model willingnesswillingness_direct_score | #46 / 156 | ↑4 | 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.