LLMAPI only
Claude 3.7 Sonnet
Anthropic
Updated May 21, 2026 · Benchmarks via Artificial Analysis, specs & pricing via OpenRouter · Methodology · Report an error
Superseded by Claude Opus 4
Specifications
- Type
- LLM
- Access
- API only
- Released
- February 24, 2025
- License
- Proprietary
- Context window
- 200,000 tokens
- Output speed
- 101 tok/s
- Latency (TTFT)
- 0.4s
- Input
- Text, Image
- Output
- Text
- API pricing
- $3.00 in · $15.00 out / 1M tokens
Capabilities
Function callingStructured output
Benchmarks
Reasoning
| GPQA Diamond | 84.8 |
Coding
| SWE-bench Verified | 70.3 |
| Terminal-Bench | 35.2 |
| LiveCodeBench | 47.3 |
Agents
Multimodal
| MMMU | 75 |
General
| MMLU | 86.1 |
| IFEval | 93.2 |
| MMLU-Pro | 83.7 |
| Humanity’s Last Exam | 10.3 |
Our take
The first hybrid reasoning model: a single model that answers instantly or, on request, thinks step by step with a visible, budgetable scratchpad. Especially strong at coding and agentic workflows, it kept Claude the developer favorite.