| Benchmark | GPT-5.1 | Kimi K2 (Reasoning) |
|---|---|---|
| AA Agentic Index | 32.2 | 47.9 |
| AA Intelligence | 37.5 | 33.5 |
| AA-LCR | 76.7 | 70.3 |
| AA-Omniscience | 5.4 | -21.4 |
| AIME 2025 | 94.2 | 94.5 |
| Artificial Analysis Coding Index | 49.4 | 34.8 |
| browsecomp | 50.8 | 60.2 |
| critpt | 4.9 | 2.6 |
| frontiermath_tier_4 | 12.5 | 0 |
| gdpval | 25 | 24.5 |
| GPQA Diamond | 88.1 | 84.5 |
| HLE | 28.5 | 23.9 |
| HLE (with tools) | 42.7 | 44.9 |
| hmmt_feb_2025 | 96.3 | 93.3 |
| hmmt_nov_2025 | 91.7 | 89.2 |
| IFBench | 72.9 | 68.1 |
| LiveCodeBench v6 | 87 | 83.1 |
| MMLU-Pro | 87 | 84.6 |
| multichallenge | 63.4 | 66.4 |
| OmniScience Accuracy | 37.7 | 30.9 |
| OmniScience Non-Hallucination | 48.1 | 25.8 |
| scicode | 43.3 | 44.8 |
| simplebench | 53.2 | 39.6 |
| swe_bench_bash | 66 | 63.4 |
| SWE-bench Verified | 76.3 | 71.3 |
| Terminal-Bench 2.0 | 47.6 | 35.7 |
| Terminal-Bench Hard | 45.5 | 31.1 |
| vectara_answer_rate | 100 | 98.6 |
| vectara_avg_summary_length | 254.4 | 59.2 |
| vectara_factual_consistency | 89.1 | 82.1 |
| vectara_hallucination_rate ↓ | 12.1 | 17.9 |
| τ²-Bench | 82.7 | 74.3 |
| τ²-Bench Telecom (AA run) | 81.9 | 93 |
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.