GPT-5.4 vs Gemini 3.1 Pro Preview
Côte à côte. Chaque métrique. Chaque benchmark.
| Type | GPT-5.4 | Gemini 3.1 Pro Preview |
|---|---|---|
| Provider |  OpenAI |  Google DeepMind |
| score moyen | 59.0 | 60.6 |
| Prix d'entrée | $2.50 | $2.00 |
| Prix de sortie | $15.00 | $12.00 |
| Fenêtre de contexte | 1.1M tokens (~525 books) | 1.0M tokens (~524 books) |
| Sorti le | 2026-03-05 | 2026-02-19 |
| Code source ouvert | Proprietary | Proprietary |
Scores de benchmark
16 benchmarks · GPT-5.4: 6, Gemini 3.1 Pro Preview: 10
| Benchmark | Catégorie | GPT-5.4 | Gemini 3.1 Pro Preview |
|---|---|---|---|
| APEX-Agents | agentic | 35.9 | 33.5 |
| ARC-AGI | reasoning | 93.7 | 98.0 |
| ARC-AGI-2 | reasoning | 74.0 | 77.1 |
| Artificial Analysis — Agentic Index | speed | 69.4 | 59.1 |
| Artificial Analysis — Coding Index | speed | 57.3 | 55.5 |
| Artificial Analysis — Quality Index | speed | 57.2 | 57.2 |
| Chatbot Arena Elo — Overall | arena | 1465.8 | 1492.6 |
| Chess Puzzles | knowledge | 44.0 | 55.0 |
| FrontierMath-2025-02-28-Private | math | 47.6 | 36.9 |
| FrontierMath-Tier-4-2025-07-01-Private | math | 27.1 | 16.7 |
| GPQA diamond | knowledge | 91.1 | 92.1 |
| OTIS Mock AIME 2024-2025 | math | 95.3 | 95.6 |
| PostTrainBench | knowledge | 20.2 | 21.6 |
| SimpleQA Verified | knowledge | 44.8 | 77.3 |
| SWE-Bench verified | coding | 76.9 | 75.6 |
| WeirdML | coding | 57.4 | 72.1 |