GPT-5.4 Pro
GPT / OpenAI
29scores
23benchmarks
$30 / $180 per 1M tokenscost in/out
Metadata
GPT Closed/API
Aliases: gpt-5.4-pro, gpt-5.4-pro-20260305, openai-gpt-5.4-pro, openai-gpt-5.4-pro-20260305, openai/gpt-5.4-pro, openai/gpt-5.4-pro-20260305
| Benchmark | Category | Rank | Score | Sampled |
|---|---|---|---|---|
| ARC-AGI-1 | Agentic | 9 | 94.50 | 2026-05-05 |
| ARC-AGI-1 | Agentic | 3 | 94.5% | 2026-04-23 |
| ARC-AGI-2 | Agentic | 6 | 83.33 | 2026-05-05 |
| ARC-AGI-2 | Agentic | 2 | 83.3% | 2026-04-23 |
| BrowseComp | Agentic | 2 | 89.3% | 2026-04-23 |
| BrowseComp | Agentic | 1 | 89.3% | 2026-04-16 |
| MultiChallenge | Agentic | 1 | 69.23 | 2026-05-06 |
| PinchBench | Agentic | 67 | 0.19 | 2026-05-06 |
| TutorBench | Education | 1 | 56.62 | 2026-05-06 |
| Vectara HHEM Hallucination Leaderboard | Factuality | 41 | 91.70 | 2026-05-06 |
| Finance Agent v1.1 | Finance | 2 | 61.5% | 2026-04-23 |
| Finance Agent v1.1 | Finance | 2 | 61.5% | 2026-04-16 |
| Investment Banking Modeling Tasks (Internal) | Finance | 4 | 83.6% | 2026-04-23 |
| TaxBench | Finance | 2 | 27.03% mean pass^5 | 2026-05-27 |
| BenchLM | General Knowledge | 4 | 91 | 2026-05-06 |
| GDPval | Generalization | 4 | 82% | 2026-04-23 |
| Humanity's Last Exam | Intelligence | 1 | 58.7% | 2026-04-23 |
| Humanity's Last Exam | Intelligence | 2 | 58.7% | 2026-04-16 |
| FrontierMath 2025-02-28 Private | Mathematics | 3 | 50% | 2026-04-23 |
| FrontierMath Tier 4 2025-07-01 Private | Mathematics | 2 | 38% | 2026-04-23 |
| Medical Chronology LLM Benchmark | Medical | 9 | 0.89 | 2026-05-06 |
| Visual-Language Understanding | Multimodal | 1 | 53.89 | 2026-05-06 |
| EnigmaEval | Reasoning | 1 | 23.82 | 2026-05-06 |
| GPQA Diamond | Reasoning | 1 | 94.4% | 2026-04-23 |
| GPQA Diamond | Reasoning | 2 | 94.4% | 2026-04-16 |
| Humanity's Last Exam (Text Only) | Reasoning | 1 | 45.32 | 2026-05-06 |
| MultiNRC | Reasoning | 1 | 62.27 | 2026-05-06 |
| CritPt | Science | 2 | 30% | 2026-05-11 |
| GeneBench | Science | 2 | 25.6% | 2026-04-23 |
No matching rows.