| Benchmark | Gemini 3.1 Pro | o4-mini |
|---|---|---|
| AA Agentic Index | 23 | 36.1 |
| AA Intelligence | 47.7 | 26.1 |
| AA-LCR | 79 | 60 |
| AA-Omniscience | 32.9 | -35.7 |
| arc_agi_1 | 98 | 58.7 |
| ARC-AGI-2 | 77.1 | 6.1 |
| arena_vision | 1280 | 1201 |
| Artificial Analysis Coding Index | 68.8 | 25.6 |
| browsecomp | 85.9 | 51.5 |
| critpt | 17.7 | 0.6 |
| Fortress | 29.8 | 21.5 |
| frontiermath_tier_4 | 16.7 | 6.3 |
| gdpval | 23.3 | 25.4 |
| GPQA Diamond | 94.3 | 81.4 |
| HLE | 51.4 | 16.5 |
| IFBench | 77.1 | 68.7 |
| LiveCodeBench | 91.7 | 84.5 |
| MMMU-Pro | 83 | 69.2 |
| multichallenge | 71.4 | 44.9 |
| MultiNRC | 64.7 | 22.2 |
| OmniScience Accuracy | 55.3 | 24.8 |
| OmniScience Non-Hallucination | 50.1 | 19.5 |
| scicode | 59 | 46.5 |
| simplebench | 79.6 | 38.7 |
| simpleqa_verified | 77.3 | 23.9 |
| strongreject | 98 | 96 |
| SWE-bench Verified | 80.6 | 68.1 |
| Terminal-Bench Hard | 68.5 | 15.2 |
| vectara_answer_rate | 99.4 | 99.2 |
| vectara_avg_summary_length | 107.7 | 130.9 |
| vectara_factual_consistency | 89.6 | 81.4 |
| vectara_hallucination_rate ↓ | 10.4 | 18.6 |
| VisualToolBench | 29 | 11.1 |
| τ²-Bench Telecom (AA run) | 99.3 | 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.