| Benchmark | GPT-5.1 | o4-mini |
|---|---|---|
| AA Agentic Index | 32.2 | 36.1 |
| AA Intelligence | 37.5 | 26.1 |
| AA-LCR | 76.7 | 60 |
| AA-Omniscience | 5.4 | -35.7 |
| AIME 2025 | 94.2 | 92.7 |
| arc_agi_1 | 72.8 | 58.7 |
| ARC-AGI-2 | 17.6 | 6.1 |
| arena_vision | 1250 | 1201 |
| Artificial Analysis Coding Index | 49.4 | 25.6 |
| browsecomp | 50.8 | 51.5 |
| critpt | 4.9 | 0.6 |
| Fortress | 25.7 | 21.5 |
| frontiermath_tier_4 | 12.5 | 6.3 |
| gdpval | 25 | 25.4 |
| GPQA Diamond | 88.1 | 81.4 |
| HLE | 28.5 | 16.5 |
| IFBench | 72.9 | 68.7 |
| MMMU | 85.4 | 81.6 |
| MMMU-Pro | 75.5 | 69.2 |
| multichallenge | 63.4 | 44.9 |
| MultiNRC | 49 | 22.2 |
| OmniScience Accuracy | 37.7 | 24.8 |
| OmniScience Non-Hallucination | 48.1 | 19.5 |
| scicode | 43.3 | 46.5 |
| SEAL VISTA | 43.8 | 51.8 |
| simplebench | 53.2 | 38.7 |
| simpleqa_verified | 48.9 | 23.9 |
| swe_bench_bash | 66 | 45 |
| SWE-bench Verified | 76.3 | 68.1 |
| Terminal-Bench Hard | 45.5 | 15.2 |
| vectara_answer_rate | 100 | 99.2 |
| vectara_avg_summary_length | 254.4 | 130.9 |
| vectara_factual_consistency | 89.1 | 81.4 |
| vectara_hallucination_rate ↓ | 12.1 | 18.6 |
| τ²-Bench Telecom (AA run) | 81.9 | 55.6 |
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.