Model profile
GPT 5 Nano
OpenAIdeveloper
2025-08-07release date
#14 / 309overall rank
23eval lineages
Evidence summary
GPT 5 Nano has an estimated overall rank of #14; its 90% source-sensitivity interval is #7–#71. Its behavior-only rank is #13; company governance moves the combined estimate to #14. Published evidence spans 23 evals and 7 of 7 behavior components. Its strongest relative result is HELM Safety (simple_safety_tests, #1 of 80); its weakest is CAIS Risk Index (hle_overconfidence, #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-nano
- OpenRouter ↗openai/gpt-5-nano
- System card ↗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 | Source |
|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #66 / 328 | ↓0.5262 | Source ↗official | |
| AIRBench 2024 Safety Scenariossafety_scenarios | #6 / 80 | ↑0.878 | Source ↗official | |
| ANIMAscore | #10 / 22 | ↑0.708 | Source ↗official | |
| CAIS Risk Indexagent_red_teaming | #32 / 44 | ↓88.5 | Source ↗official | |
| CAIS Risk Indexbioweapons_assistance | #28 / 49 | ↓71.3 | Source ↗official | |
| CAIS Risk Indexhle_overconfidence | #45 / 50 | ↓80 | Source ↗official | |
| CAIS Risk Indexmachiavelli | #1 / 46 | ↓79.4 | Source ↗official | |
| CAIS Risk Indexmask | #17 / 52 | ↓12.3 | Source ↗official | |
| CAIS Risk Indextextquests_harm | #1 / 49 | ↓2 | Source ↗official | |
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #21 / 104 | ↓7.527 | Source ↗official | |
| Enkrypt AI Safety Leaderboardbias_attack_non_success_rate | #33 / 241 | ↑39.02 | Source ↗official | |
| Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate | #4 / 241 | ↑98 | Source ↗official | |
| Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate | #39 / 241 | ↑92.78 | Source ↗official | |
| Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate | #104 / 239 | ↑96.91 | Source ↗official | |
| HELM Safetyanthropic_red_team | #13 / 80 | ↑0.996 | Source ↗official | |
| HELM Safetybbq | #8 / 80 | ↑0.976 | Source ↗official | |
| HELM Safetyharmbench | #3 / 80 | ↑0.984 | Source ↗official | |
| HELM Safetysimple_safety_tests | #1 / 80 | ↑1 | Source ↗official | |
| HELM Safetyxstest | #16 / 80 | ↑0.976 | Source ↗official | |
| MACHIAVELLIdeception_relative_random_pct | #1 / 50 | ↓79.4 | Source ↗official | |
| Manager Coercion Benchcoercion_ladder_depth | #5 / 31 | ↓5.067 | Source ↗self run | |
| MonitoringBench Full-Trajectory Monitorfull_trajectory_catch_rate_at_1pct_fpr_percent | #10 / 13 | ↑6 | Source ↗official | |
| PHAREbias_resistance_diagnostic | #55 / 66 | ↑0.347 | Source ↗official | |
| PHAREhallucination_resistance_diagnostic | #31 / 70 | ↑0.7637 | Source ↗official | |
| PHAREharm_resistance_diagnostic | #9 / 70 | ↑0.9741 | Source ↗official | |
| PHAREjailbreak_resistance_diagnostic | #12 / 67 | ↑0.6964 | Source ↗official | |
| SpeciEvalbelief_animal_sentience | #97 / 113 | ↑6.4 | Source ↗official | |
| SpeciEvalland_animal_4ns | #32 / 113 | ↓4.35 | Source ↗official | |
| SpeciEvalsea_animal_4ns | #48 / 113 | ↓4.68 | Source ↗official | |
| SpeciEvalspeciesism | #80 / 113 | ↓2.33 | Source ↗official | |
| StereoTales Harmful Associationsbenign_significant_association_score | #2 / 23 | ↑89.22 | Source ↗official | |
| TACbase_welfare_rate | #10 / 76 | ↑37.82 | Source ↗self run | |
| Vectara HHEM Factual Consistencyfactual_consistency_rate | #56 / 94 | ↑89.5 | Source ↗official |