o3 vs Gemini 3.1 Pro Preview
Lado a lado. Cada métrica. Cada benchmark.
| Tipo | o3 | Gemini 3.1 Pro Preview |
|---|---|---|
| Provider |  OpenAI |  Google DeepMind |
| pontuação média | 55.2 | 60.6 |
| Preço de entrada | $2.00 | $2.00 |
| Preço de saída | $8.00 | $12.00 |
| Janela de contexto | 200K tokens (~100 books) | 1.0M tokens (~524 books) |
| Lançado em | 2025-04-16 | 2026-02-19 |
| Código aberto | Proprietary | Proprietary |
Pontuações de benchmark
14 benchmarks · o3: 0, Gemini 3.1 Pro Preview: 14
| Benchmark | Categoria | o3 | Gemini 3.1 Pro Preview |
|---|---|---|---|
| ARC-AGI | reasoning | 60.8 | 98.0 |
| ARC-AGI-2 | reasoning | 6.5 | 77.1 |
| Artificial Analysis — Agentic Index | speed | 36.1 | 59.1 |
| Artificial Analysis — Coding Index | speed | 38.4 | 55.5 |
| Artificial Analysis — Quality Index | speed | 38.4 | 57.2 |
| EnigmaEval | knowledge | 13.1 | 19.8 |
| FrontierMath-2025-02-28-Private | math | 18.7 | 36.9 |
| FrontierMath-Tier-4-2025-07-01-Private | math | 2.1 | 16.7 |
| GPQA diamond | knowledge | 75.8 | 92.1 |
| OTIS Mock AIME 2024-2025 | math | 83.9 | 95.6 |
| SimpleBench | reasoning | 43.7 | 75.5 |
| SimpleQA Verified | knowledge | 53.0 | 77.3 |
| SWE-Bench verified | coding | 62.3 | 75.6 |
| WeirdML | coding | 52.4 | 72.1 |