| Benchmark | Claude Opus 4.6 | Kimi K2 (Reasoning) |
|---|---|---|
| AA Agentic Index | 67.6 | 47.9 |
| AA Intelligence | 44.9 | 33.5 |
| AA-LCR | 74.3 | 70.3 |
| AA-Omniscience | 13.7 | -21.4 |
| AIME 2025 | 99.8 | 94.5 |
| AIME25 no tools | 95.6 | 94.5 |
| Artificial Analysis Coding Index | 48.1 | 34.8 |
| browsecomp | 86.8 | 60.2 |
| browsecomp_with_context_manager | 84 | 60.2 |
| critpt | 12.6 | 2.6 |
| frontiermath_tier_4 | 22.9 | 0 |
| gdpval | 55.9 | 24.5 |
| GPQA Diamond | 91.3 | 84.5 |
| HLE | 53.1 | 23.9 |
| HLE (with tools) | 62.7 | 44.9 |
| hmmt_nov_2025 | 96.3 | 89.2 |
| IFBench | 62.5 | 68.1 |
| imo_answer_bench | 75.3 | 78.6 |
| LiveCodeBench | 88.8 | 79.2 |
| LiveCodeBench v6 | 88.8 | 83.1 |
| MMLU-Pro | 89.1 | 84.6 |
| multichallenge | 56 | 66.4 |
| OmniScience Accuracy | 47 | 30.9 |
| OmniScience Non-Hallucination | 37.2 | 25.8 |
| scicode | 52 | 44.8 |
| simplebench | 67.6 | 39.6 |
| swe_bench_bash | 75.6 | 63.4 |
| SWE-bench Multilingual | 77.8 | 61.1 |
| SWE-bench Verified | 80.8 | 71.3 |
| Terminal-Bench 2.0 | 65.4 | 35.7 |
| Terminal-Bench Hard | 65.4 | 31.1 |
| vectara_answer_rate | 99.8 | 98.6 |
| vectara_avg_summary_length | 137.6 | 59.2 |
| vectara_factual_consistency | 87.8 | 82.1 |
| vectara_hallucination_rate ↓ | 12.2 | 17.9 |
| τ²-Bench Telecom (AA run) | 99.3 | 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.