Claude 3.7 Sonnet
来自 Anthropic · 发布于 2025-02-24
47.7
平均分
$3.00/1M
输入价格
$15.00/1M
输出价格
200K tokens (~100 books)
上下文窗口
multimodal
类型
Tested on 26 benchmarks with 47.7% average. Top scores: MATH level 5 (91.2%), HELM — IFEval (83.4%), Fiction.LiveBench (83.3%).
基准测试分数
| 基准测试 | 类别 | 分数 | Bar |
|---|---|---|---|
| MATH level 5 | math | 91.2 | |
| HELM — IFEval | language | 83.4 | |
| Fiction.LiveBench | knowledge | 83.3 | |
| HELM — WildBench | reasoning | 81.4 | |
| Lech Mazur Writing | knowledge | 81.1 | |
| HELM — MMLU-Pro | knowledge | 78.4 | |
| GPQA diamond | knowledge | 73.0 | |
| GeoBench | knowledge | 68.0 | |
| Aider polyglot | coding | 64.9 | |
| SWE-Bench verified | coding | 61.0 | |
| HELM — GPQA | knowledge | 60.8 | |
| OTIS Mock AIME 2024-2025 | math | 57.7 | |
| CadEval | coding | 54.0 | |
| SWE-Bench Verified (Bash Only) | coding | 52.8 | |
| DeepResearch Bench | knowledge | 43.6 | |
| OSWorld | agentic | 35.8 | |
| SimpleBench | reasoning | 35.7 | |
| HELM — Omni-MATH | math | 33.0 | |
| The Agent Company | agentic | 30.9 | |
| ARC-AGI | reasoning | 28.6 | |
| Cybench | coding | 20.0 | |
| VPCT | knowledge | 8.5 | |
| FrontierMath-2025-02-28-Private | math | 4.1 | |
| GSO-Bench | coding | 3.8 | |
| HLE | knowledge | 3.4 | |
| ARC-AGI-2 | reasoning | 0.9 |
相似模型
U
Stable Beluga 2Unknown
47.8
Anthropic
47.6
Google DeepMind
48.0
Google DeepMind
47.4