Claude Opus 4.5 vs Gemini 3.1 Pro Preview
Lado a lado. Cada métrica. Cada benchmark.
| Tipo | Claude Opus 4.5 | Gemini 3.1 Pro Preview |
|---|---|---|
| Provider |  Anthropic |  Google DeepMind |
| pontuação média | 45.4 | 60.6 |
| Preço de entrada | $5.00 | $2.00 |
| Preço de saída | $25.00 | $12.00 |
| Janela de contexto | 200K tokens (~100 books) | 1.0M tokens (~524 books) |
| Lançado em | 2025-11-24 | 2026-02-19 |
| Código aberto | Proprietary | Proprietary |
Pontuações de benchmark
16 benchmarks · Claude Opus 4.5: 2, Gemini 3.1 Pro Preview: 14
| Benchmark | Categoria | Claude Opus 4.5 | Gemini 3.1 Pro Preview |
|---|---|---|---|
| APEX-Agents | agentic | 18.4 | 33.5 |
| ARC-AGI | reasoning | 80.0 | 98.0 |
| ARC-AGI-2 | reasoning | 37.6 | 77.1 |
| Chatbot Arena Elo — Coding | arena | 1465.2 | 1455.7 |
| Chatbot Arena Elo — Overall | arena | 1467.7 | 1492.6 |
| Chess Puzzles | knowledge | 12.0 | 55.0 |
| FrontierMath-2025-02-28-Private | math | 20.7 | 36.9 |
| FrontierMath-Tier-4-2025-07-01-Private | math | 4.2 | 16.7 |
| GPQA diamond | knowledge | 81.4 | 92.1 |
| OTIS Mock AIME 2024-2025 | math | 86.1 | 95.6 |
| PostTrainBench | knowledge | 17.3 | 21.6 |
| SimpleBench | reasoning | 54.4 | 75.5 |
| SimpleQA Verified | knowledge | 41.8 | 77.3 |
| SWE-Bench verified | coding | 76.7 | 75.6 |
| Terminal Bench | coding | 63.1 | 78.4 |
| WeirdML | coding | 63.7 | 72.1 |