| Benchmark | GPT-4o | GPT-5.1 |
|---|---|---|
| AA Agentic Index | 9.7 | 32.2 |
| AA Intelligence | 11.2 | 37.5 |
| AA-LCR | 55 | 76.7 |
| AA-Omniscience | -10.5 | 5.4 |
| AIME 2025 | 11.7 | 94.2 |
| arc_agi_1 | 4.5 | 72.8 |
| ARC-AGI-2 | 0 | 17.6 |
| arena_vision | 1162 | 1250 |
| Artificial Analysis Coding Index | 24.2 | 49.4 |
| critpt | 0 | 4.9 |
| Fortress | 47.2 | 25.7 |
| gdpval | 0 | 25 |
| GPQA Diamond | 70.1 | 88.1 |
| HLE | 5.3 | 28.5 |
| IFBench | 36 | 72.9 |
| LiveCodeBench v6 | 30.9 | 87 |
| MMLU-Pro | 74.7 | 87 |
| mmmlu | 81.4 | 91 |
| MMMU | 72.2 | 85.4 |
| MMMU-Pro | 59.9 | 75.5 |
| multichallenge | 40.3 | 63.4 |
| MultiNRC | 12.4 | 49 |
| OmniScience Accuracy | 23.7 | 37.7 |
| OmniScience Non-Hallucination | 62.1 | 48.1 |
| scicode | 33.4 | 43.3 |
| SEAL VISTA | 34.9 | 43.8 |
| simplebench | 17.8 | 53.2 |
| swe_bench_bash | 21.6 | 66 |
| SWE-bench Verified | 38.8 | 76.3 |
| Tau2 airline | 45.5 | 67 |
| Tau2 retail | 63.4 | 77.9 |
| Terminal-Bench Hard | 8.3 | 45.5 |
| vectara_answer_rate | 93.8 | 100 |
| vectara_avg_summary_length | 86.6 | 254.4 |
| vectara_factual_consistency | 90.4 | 89.1 |
| vectara_hallucination_rate ↓ | 9.6 | 12.1 |
| τ²-Bench Telecom (AA run) | 28.9 | 81.9 |
Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.