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