Model profile
OLMoE 1B 7B Instruct
Evidence summary
Safety. OLMoE 1B 7B Instruct has an estimated Safety rank of #337; its 90% source-sensitivity interval is #263–#340. Its behavior-only rank is #346; company governance moves the combined estimate to #337. Published Safety evidence spans 6 eval lineages and 5 of 7 components. Its strongest relative result is Enkrypt AI Safety Leaderboard (bias_attack_non_success_rate, #152 of 270); its weakest is HELM Safety (bbq, #80 of 80).
Freedom. OLMoE 1B 7B Instruct has an estimated Freedom rank of #69; its 90% source-sensitivity interval is #8–#542. Published Freedom evidence spans 5 eval lineages and 1 of 1 components. Its strongest relative result is Enkrypt AI Safety Leaderboard (toxicity_attack_non_success_rate, #7 of 268); its weakest is HELM Safety (xstest, #76 of 80).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Hugging Face ↗allenai/OLMoE-1B-7B-0125-Instruct
- 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 |
|---|---|---|---|---|
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #152 / 270 | ↑14.99 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #171 / 270 | ↑84.5 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #258 / 270 | ↑31.11 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #262 / 268 | ↑64.09 | Source ↗official | |
| HELM Safetyanthropic_red_team | #77 / 80 | ↑0.889 | Source ↗official | |
| HELM Safetybbq | #80 / 80 | ↑0.419 | Source ↗official | |
| HELM Safetyharmbench | #58 / 80 | ↑0.629 | Source ↗official | |
| HELM Safetysimple_safety_tests | #78 / 80 | ↑0.725 | Source ↗official | |
| HELM Safetyxstest | #76 / 80 | ↑0.843 | 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 |
|---|---|---|---|---|
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #100 / 270 | ↓84.5 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #12 / 270 | ↓31.11 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #7 / 268 | ↓64.09 | Source ↗official | |
| HELM Safetyanthropic_red_team | #4 / 80 | ↓0.889 | Source ↗official | |
| HELM Safetyharmbench | #23 / 80 | ↓0.629 | Source ↗official | |
| HELM Safetysimple_safety_tests | #3 / 80 | ↓0.725 | Source ↗official | |
| HELM Safetyxstest | #76 / 80 | ↑0.843 | Source ↗official |