Model profile
GPT 5 Pro
Evidence summary
Safety. GPT 5 Pro has an estimated Safety rank of #45; its 90% source-sensitivity interval is #4–#278. Its behavior-only rank is #46; company governance moves the combined estimate to #45. Published Safety evidence spans 5 eval lineages and 4 of 7 components. Its strongest relative result is SpeciEval (sea_animal_4ns, #10 of 131); its weakest is FORTRESS (over_refusal_score, #52 of 59).
Freedom. GPT 5 Pro has an estimated Freedom rank of #648; its 90% source-sensitivity interval is #343–#661. Published Freedom evidence spans 2 eval lineages and 1 of 1 components. Its strongest relative result is FORTRESS (average_risk_score, #50 of 60); its weakest is FORTRESS (over_refusal_score, #52 of 59).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- OpenRouter ↗openai/gpt-5-pro
- Official model documentation ↗Family-level model document · openai · first party
- Release source ↗curated outward evidence
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 |
|---|---|---|---|---|
| FORTRESSaverage_risk_score | #11 / 60 | ↓15.2 | Source ↗official | |
| FORTRESSover_refusal_score | #52 / 59 | ↓12.43 | Source ↗official | |
| Human Pathogen Capabilities Test (HPCT) — overall refusalrefusal_rate_pct | #12 / 69 | ↑46.9 | Source ↗official | |
| Humanity's Last Exam RMS calibration error (Scale Labs)calibrationError | #9 / 42 | ↓49 | Source ↗official | |
| MASKlying_probability_pct | #11 / 53 | ↓14.01 | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #58 / 131 | ↑6.83 | Source ↗official | |
| SpeciEvalland_animal_4ns | #22 / 131 | ↓4.25 | Source ↗official | |
| SpeciEvalsea_animal_4ns | #10 / 131 | ↓4.25 | Source ↗official | |
| SpeciEvalspeciesism | #16 / 131 | ↓1.45 | 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.