Model profile
Granite 3.3 8B Instruct
Evidence summary
Safety. Granite 3.3 8B Instruct has an estimated Safety rank of #284; its 90% source-sensitivity interval is #67–#315. Its behavior-only rank is #288; company governance moves the combined estimate to #284. Published Safety evidence spans 9 eval lineages and 4 of 7 components. Its strongest relative result is BlueBench AttaQ-100 (attaq_harmlessness_reward_pct, #1 of 18); its weakest is AA-Omniscience (hallucination_rate, #337 of 345).
Freedom. Granite 3.3 8B Instruct has an estimated Freedom rank of #492; its 90% source-sensitivity interval is #343–#581. Published Freedom evidence spans 6 eval lineages and 1 of 1 components. Its strongest relative result is HELM Safety (simple_safety_tests, #24 of 80); its weakest is BlueBench AttaQ-100 (attaq_harmlessness_reward_pct, #18 of 18).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗granite-3-3-8b-instruct
- Hugging Face ↗ibm-granite/granite-3.3-8b-instruct
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗direct research preview identity date
Finetunes
Direct finetunes linked to this canonical base model. Quantized and repackaged derivatives are excluded.
- heterodoxin/granite-3.3-8b-instruct-apostate ↗Direct non-quantized finetune documented by Hugging Face metadata
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 |
|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #337 / 345 | ↓0.9622 | Source ↗official | |
| AIRBench 2024 Safety Scenariossafety_scenarios | #25 / 80 | ↑0.76 | Source ↗official | |
| BlueBench AttaQ-100attaq_harmlessness_reward_pct | #1 / 18 | ↑88.44 | Source ↗official | |
| HELM Safetyanthropic_red_team | #32 / 80 | ↑0.991 | Source ↗official | |
| HELM Safetybbq | #65 / 80 | ↑0.84 | Source ↗official | |
| HELM Safetyharmbench | #34 / 80 | ↑0.812 | Source ↗official | |
| HELM Safetysimple_safety_tests | #54 / 80 | ↑0.975 | Source ↗official | |
| HELM Safetyxstest | #68 / 80 | ↑0.907 | Source ↗official | |
| Vectara HHEM Factual Consistencyfactual_consistency_rate | #59 / 94 | ↑89.4 | 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 | #56 / 80 | ↓0.76 | Source ↗official | |
| BlueBench AttaQ-100attaq_harmlessness_reward_pct | #18 / 18 | ↓88.44 | Source ↗official | |
| HELM Safetyanthropic_red_team | #45 / 80 | ↓0.991 | Source ↗official | |
| HELM Safetyharmbench | #47 / 80 | ↓0.812 | Source ↗official | |
| HELM Safetysimple_safety_tests | #24 / 80 | ↓0.975 | Source ↗official | |
| HELM Safetyxstest | #68 / 80 | ↑0.907 | Source ↗official |