Claude Sonnet 4 vs Llama 4 Maverick
Lado a lado. Cada métrica. Cada benchmark.
| Tipo | Claude Sonnet 4 | Llama 4 Maverick |
|---|---|---|
| Provider |  Anthropic |  Meta |
| pontuação média | 44.6 | 28.0 |
| Preço de entrada | $3.00 | $0.15 |
| Preço de saída | $15.00 | $0.60 |
| Janela de contexto | 1.0M tokens (~500 books) | 1.0M tokens (~524 books) |
| Lançado em | 2025-05-22 | 2025-04-05 |
| Código aberto | Proprietary | Open Source |
Pontuações de benchmark
13 benchmarks · Claude Sonnet 4: 12, Llama 4 Maverick: 1
| Benchmark | Categoria | Claude Sonnet 4 | Llama 4 Maverick |
|---|---|---|---|
| Aider polyglot | coding | 61.3 | 15.6 |
| ARC-AGI | reasoning | 40.0 | 4.4 |
| ARC-AGI-2 | reasoning | 5.9 | 0.1 |
| Fiction.LiveBench | knowledge | 46.9 | 46.2 |
| FrontierMath-2025-02-28-Private | math | 4.1 | 0.7 |
| GeoBench | knowledge | 37.0 | 52.0 |
| GPQA diamond | knowledge | 72.3 | 56.0 |
| HLE | knowledge | 3.1 | 0.9 |
| MATH level 5 | math | 84.4 | 73.0 |
| OTIS Mock AIME 2024-2025 | math | 71.1 | 20.5 |
| SimpleBench | reasoning | 34.6 | 13.2 |
| SWE-Bench Verified (Bash Only) | coding | 64.9 | 21.0 |
| WeirdML | coding | 46.1 | 24.5 |