Gemini 3 Pro (Preview) (High): benchmark results
Gemini 3 Pro (Preview) evaluated at the high reasoning-effort setting. Provider: Google. Released 2025-11-18. Access: API.
Aggregate Rating 1802 ± 11, rank #136 of 2168 rated models, from 149 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA LiveCodeBench | 91.75 | Pass@1 (%) | 100 |
| AA MMLU-Pro | 89.76 | Accuracy (%) | 100 |
| AA Omniscience - Software Engineering (SWE) - Java | 55 | Accuracy (%) | 100 |
| Agent Planning Benchmark - Step-wise Planning | 80.1 | Correctness rate (%) of the next 1 to 3 actions predicted fr | 100 |
| Agent Planning Benchmark - Tool-Broken Recovery | 78 | Replace rate (%): share of 300 cases in which, after a criti | 100 |
| DocVQA 2026 | 37.5 | Accuracy (%, high thinking) | 100 |
| DocVQA 2026 - Infographics | 70 | Accuracy (%, high thinking) | 100 |
| DocVQA 2026 - Science Poster | 30 | Accuracy (%, high thinking) | 100 |
| DocVQA 2026 - Slide | 70 | Accuracy (%, high thinking) | 100 |
| Eleusis Benchmark | 17.26 | Avg Floored Score | 100 |
| Vals AI MATH 500 | 96.4 | Accuracy (%) | 100 |
| AA Omniscience - Software Engineering (SWE) - JavaScript | 75.45 | Accuracy (%) | 99.7 |
Interactive version: theaggregate.ai/model?slug=gemini-3-pro-preview-high · How It Works · Data refreshed daily, snapshot 2026-10-20.