Model profile
GPT 5.4 Pro
Evidence summary
Safety. GPT 5.4 Pro has an estimated Safety rank of #60; its 90% source-sensitivity interval is #1–#275. Its behavior-only rank is #70; company governance moves the combined estimate to #60. Published Safety evidence spans 4 eval lineages and 3 of 7 components. Its strongest relative result is MASK (lying_probability_pct, #3 of 53); its weakest is FORTRESS (over_refusal_score, #38 of 49).
Freedom. GPT 5.4 Pro has an estimated Freedom rank of #600; its 90% source-sensitivity interval is #339–#636. Published Freedom evidence spans 1 eval lineages and 1 of 1 components. Its strongest relative result is FORTRESS (over_refusal_score, #38 of 49); its weakest is FORTRESS (average_risk_score, #45 of 50).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗gpt-5-4-pro
- OpenRouter ↗openai/gpt-5.4-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 | #6 / 50 | ↓14.84 | Source ↗official | |
| FORTRESSover_refusal_score | #38 / 49 | ↓9.41 | Source ↗official | |
| Humanity's Last Exam RMS calibration error (Scale Labs)calibrationError | #3 / 42 | ↓38 | Source ↗official | |
| MASKlying_probability_pct | #3 / 53 | ↓8.27 | Source ↗official | |
| Vectara HHEM Factual Consistencyfactual_consistency_rate | #38 / 94 | ↑91.7 | 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.