| Benchmark | Gemini 3.1 Pro | Kimi K2 (Reasoning) |
|---|---|---|
| AA Agentic Index | 23 | 47.9 |
| AA Intelligence | 47.7 | 33.5 |
| AA-LCR | 79 | 70.3 |
| AA-Omniscience | 32.9 | -21.4 |
| Artificial Analysis Coding Index | 68.8 | 34.8 |
| browsecomp | 85.9 | 60.2 |
| browsecomp_with_context_manager | 85.9 | 60.2 |
| critpt | 17.7 | 2.6 |
| frontiermath_tier_4 | 16.7 | 0 |
| gdpval | 23.3 | 24.5 |
| GPQA Diamond | 94.3 | 84.5 |
| HLE | 51.4 | 23.9 |
| HLE (with tools) | 51.6 | 44.9 |
| hmmt_nov_2025 | 94.8 | 89.2 |
| IFBench | 77.1 | 68.1 |
| imo_answer_bench | 91 | 78.6 |
| LiveCodeBench | 91.7 | 79.2 |
| LiveCodeBench v6 | 91.7 | 83.1 |
| MMLU-Pro | 91 | 84.6 |
| multichallenge | 71.4 | 66.4 |
| OmniScience Accuracy | 55.3 | 30.9 |
| OmniScience Non-Hallucination | 50.1 | 25.8 |
| scicode | 59 | 44.8 |
| simplebench | 79.6 | 39.6 |
| SWE-bench Multilingual | 76.9 | 61.1 |
| SWE-bench Verified | 80.6 | 71.3 |
| Terminal-Bench 2.0 | 68.5 | 35.7 |
| Terminal-Bench Hard | 68.5 | 31.1 |
| vectara_answer_rate | 99.4 | 98.6 |
| vectara_avg_summary_length | 107.7 | 59.2 |
| vectara_factual_consistency | 89.6 | 82.1 |
| vectara_hallucination_rate ↓ | 10.4 | 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.