Maintenance update to GPT-5 with steerability + latency improvements.
Major GPT-5 iteration with adaptive reasoning and perfect scores on math competitions
| Benchmark | GPT-5.1 | GPT-5 | Δ |
|---|---|---|---|
| ARC-AGI | 87.5% | — | — |
| AIME 2025 | 100% | — | — |
| MMLU | 92.5% | 91.0% | +1.5 |
| MMLU-Pro | 89.2% | 87.5% | +1.7 |
| GPQA Diamond | 77.8% | 74.2% | +3.6 |
| SWE-bench Verified | 70.1% | 67.4% | +2.7 |