| Benchmark | Gemini 3 Flash Preview | Kimi K2 (Reasoning) |
|---|---|---|
| AA Agentic Index | 49.7 | 47.9 |
| AA Intelligence | 38.7 | 33.5 |
| AA-LCR | 73 | 70.3 |
| AA-Omniscience | 10.1 | -21.4 |
| AIME 2025 | 99.7 | 94.5 |
| Artificial Analysis Coding Index | 42.6 | 34.8 |
| critpt | 8.6 | 2.6 |
| frontiermath_tier_4 | 4.2 | 0 |
| gdpval | 35.2 | 24.5 |
| GPQA Diamond | 90.4 | 84.5 |
| HLE | 43.5 | 23.9 |
| hmmt_feb_2025 | 97.5 | 93.3 |
| hmmt_nov_2025 | 93.3 | 89.2 |
| IFBench | 78 | 68.1 |
| OmniScience Accuracy | 53.4 | 30.9 |
| OmniScience Non-Hallucination | 7.6 | 25.8 |
| scicode | 50.6 | 44.8 |
| simplebench | 61.1 | 39.6 |
| swe_bench_bash | 75.8 | 63.4 |
| SWE-bench Multilingual | 72.7 | 61.1 |
| SWE-bench Verified | 78 | 71.3 |
| Terminal-Bench 2.0 | 47.6 | 35.7 |
| Terminal-Bench Hard | 38.6 | 31.1 |
| vectara_answer_rate | 99.8 | 98.6 |
| vectara_avg_summary_length | 90.2 | 59.2 |
| vectara_factual_consistency | 86.5 | 82.1 |
| vectara_hallucination_rate ↓ | 13.5 | 17.9 |
| τ²-Bench Telecom (AA run) | 80.4 | 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.