Model profile
Zephyr 7B Beta
Evidence summary
Zephyr 7B Beta has an estimated overall rank of #233; its 90% source-sensitivity interval is #96–#257. Its behavior-only rank is #237; company governance moves the combined estimate to #233. Published evidence spans 5 evals and 4 of 7 behavior components. Its strongest relative result is Enkrypt AI Safety Leaderboard (cbrn_attack_non_success_rate, #71 of 260); its weakest is HarmBench (dr, #28 of 28).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Hugging Face ↗HuggingFaceH4/zephyr-7b-beta
- OpenRouter ↗huggingfaceh4/zephyr-7b-beta
- 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 |
|---|---|---|---|---|---|
| Contextual MoralChoicehuman_agreement | #12 / 22 | 0.4 | ↑ higher | Source ↗official | |
| DecodingTrustmachine_ethics | #6 / 8 | 41.03 | ↑ higher | Source ↗official | |
| DecodingTruststereotype_bias | #5 / 8 | 92.6 | ↑ higher | Source ↗official | |
| DecodingTrusttoxicity | #7 / 8 | 31.97 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #134 / 260 | 15.5 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #71 / 260 | 91.17 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #259 / 260 | 20 | ↑ higher | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #182 / 258 | 93.94 | ↑ higher | Source ↗official | |
| HarmBenchdr | #28 / 28 | 65.8 | ↓ lower | Source ↗official | |
| SORRY-Benchavg | #50 / 51 | 0.87 | ↓ lower | Source ↗official |
