Gemini 2.5 Pro vs o3
Côte à côte. Chaque métrique. Chaque benchmark.
| Type | Gemini 2.5 Pro | o3 |
|---|---|---|
| Provider |  Google DeepMind |  OpenAI |
| score moyen | 56.2 | 55.2 |
| Prix d'entrée | $1.25 | $2.00 |
| Prix de sortie | $10.00 | $8.00 |
| Fenêtre de contexte | 1.0M tokens (~524 books) | 200K tokens (~100 books) |
| Sorti le | 2025-06-17 | 2025-04-16 |
| Code source ouvert | Proprietary | Proprietary |
Scores de benchmark
28 benchmarks · Gemini 2.5 Pro: 13, o3: 15
| Benchmark | Catégorie | Gemini 2.5 Pro | o3 |
|---|---|---|---|
| Aider polyglot | coding | 83.1 | 81.3 |
| ARC-AGI | reasoning | 41.0 | 60.8 |
| ARC-AGI-2 | reasoning | 4.9 | 6.5 |
| Artificial Analysis — Agentic Index | speed | 32.7 | 36.1 |
| Artificial Analysis — Coding Index | speed | 31.9 | 38.4 |
| Artificial Analysis — Quality Index | speed | 34.6 | 38.4 |
| CadEval | coding | 64.0 | 74.0 |
| DeepResearch Bench | knowledge | 49.7 | 46.6 |
| Fiction.LiveBench | knowledge | 91.7 | 88.9 |
| FrontierMath-2025-02-28-Private | math | 14.1 | 18.7 |
| FrontierMath-Tier-4-2025-07-01-Private | math | 4.2 | 2.1 |
| GeoBench | knowledge | 81.0 | 74.0 |
| GPQA diamond | knowledge | 80.4 | 75.8 |
| GSO-Bench | coding | 3.9 | 8.8 |
| HELM — GPQA | knowledge | 74.9 | 75.3 |
| HELM — IFEval | language | 84.0 | 86.9 |
| HELM — MMLU-Pro | knowledge | 86.3 | 85.9 |
| HELM — Omni-MATH | math | 41.6 | 71.4 |
| HELM — WildBench | reasoning | 85.7 | 86.1 |
| HLE | knowledge | 17.7 | 16.3 |
| Lech Mazur Writing | knowledge | 86.0 | 83.9 |
| MATH level 5 | math | 95.6 | 97.8 |
| OTIS Mock AIME 2024-2025 | math | 84.7 | 83.9 |
| SimpleBench | reasoning | 54.9 | 43.7 |
| SimpleQA Verified | knowledge | 56.0 | 53.0 |
| SWE-Bench verified | coding | 57.6 | 62.3 |
| VPCT | knowledge | 19.6 | 28.0 |
| WeirdML | coding | 54.0 | 52.4 |