Claude Opus 4.5 vs Gemini 3 Flash Preview
Lado a lado. Cada métrica. Cada benchmark.
| Tipo | Claude Opus 4.5 | Gemini 3 Flash Preview |
|---|---|---|
| Provider |  Anthropic |  Google DeepMind |
| pontuação média | 45.4 | 49.1 |
| Preço de entrada | $5.00 | $0.50 |
| Preço de saída | $25.00 | $3.00 |
| Janela de contexto | 200K tokens (~100 books) | 1.0M tokens (~524 books) |
| Lançado em | 2025-11-24 | 2025-12-17 |
| Código aberto | Proprietary | Proprietary |
Pontuações de benchmark
20 benchmarks · Claude Opus 4.5: 10, Gemini 3 Flash Preview: 9
| Benchmark | Categoria | Claude Opus 4.5 | Gemini 3 Flash Preview |
|---|---|---|---|
| APEX-Agents | agentic | 18.4 | 24.0 |
| ARC-AGI | reasoning | 80.0 | 21.5 |
| ARC-AGI-2 | reasoning | 37.6 | 33.6 |
| Chatbot Arena Elo — Coding | arena | 1465.2 | 1436.4 |
| Chatbot Arena Elo — Overall | arena | 1467.7 | 1473.9 |
| Chess Puzzles | knowledge | 12.0 | 38.0 |
| FrontierMath-2025-02-28-Private | math | 20.7 | 35.6 |
| FrontierMath-Tier-4-2025-07-01-Private | math | 4.2 | 4.2 |
| GeoBench | knowledge | 75.0 | 88.0 |
| GPQA diamond | knowledge | 81.4 | 77.6 |
| GSO-Bench | coding | 26.5 | 9.8 |
| MCP Atlas | agentic | 62.3 | 57.4 |
| OTIS Mock AIME 2024-2025 | math | 86.1 | 92.8 |
| SciPredict | knowledge | 23.1 | 22.2 |
| SimpleBench | reasoning | 54.4 | 53.3 |
| SimpleQA Verified | knowledge | 41.8 | 67.4 |
| SWE-Bench verified | coding | 76.7 | 75.4 |
| Terminal Bench | coding | 63.1 | 64.3 |
| VPCT | knowledge | 10.0 | 58.9 |
| WeirdML | coding | 63.7 | 61.6 |