Model profile
Davinci
OpenAIdeveloper
2020-06-11release date
#246 / 267overall rank
3eval lineages
Evidence summary
Davinci has an estimated overall rank of #246; its 90% source-sensitivity interval is #188–#256. Its behavior-only rank is #256; company governance moves the combined estimate to #246. Published evidence spans 3 evals and 4 of 7 behavior components. Its strongest relative result is SafetyPrompts (score, #12 of 15); its weakest is HELM Classic RealToxicityPrompts (non_toxic_prompt_toxic_fraction, #42 of 42).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Official model documentation ↗Family-level model document · openai · first party
- 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 |
|---|---|---|---|---|---|
| HELM Classic RealToxicityPromptsnon_toxic_prompt_toxic_fraction | #42 / 42 | 0.03449 | ↓ lower | Source ↗official | |
| HELM Classic RealToxicityPromptstoxic_prompt_toxic_fraction | #37 / 42 | 0.1049 | ↓ lower | Source ↗official | |
| MACHIAVELLIdeception_relative_random_pct | #47 / 50 | 105 | ↓ lower | Source ↗official | |
| SafetyPromptsscore | #12 / 15 | 42.98 | ↑ higher | Source ↗official |