Overview
Claude 3.7 Sonnet was released on February 24, 2025, and was Anthropic's first hybrid reasoning model. A single model could either answer immediately or engage extended thinking to work through a problem step by step, with API users able to set a thinking budget of up to 128K tokens. It launched alongside the research preview of Claude Code.
On SWE-bench Verified, Claude 3.7 Sonnet reached 62.3% in standard mode and 70.3% with extended thinking and additional scaffolding, leading the field on agentic coding at the time. Anthropic also reported it reduced unnecessary refusals by 45% compared with Claude 3.5 Sonnet, making it more useful on borderline requests.
Claude 3.7 Sonnet has a 200K token context window and up to 128K output tokens, and supports text and vision input, at $3 / $15 per million tokens (output pricing includes thinking tokens). The dated snapshot claude-3-7-sonnet-20250219 was retired on February 19, 2026; Anthropic recommends Sonnet 4.6 as the replacement.
| Released | 2025-02-24 |
|---|---|
| License | Proprietary |
| Weights | API only |
| Context | 200K |
| Max output | 128K |
| Architecture | Proprietary transformer; first Anthropic hybrid reasoning model with toggleable extended thinking (budget up to 128K tokens). |
| Knowledge cutoff | November 2024 |
| Modalities | Text, Vision, PDF |
| Status | Deprecated |
Benchmarks
Scores on a 0–100 scale (25-point gridlines); higher is better. Each benchmark links to its published source.
Pricing
| Input | $3.00 per million tokens |
|---|---|
| Output | $15.00 per million tokens |
Output price includes thinking (reasoning) tokens.
Strengths
- First Claude model with toggleable extended thinking in one model
- Configurable thinking budget up to 128K tokens via the API
- Leading agentic-coding performance at launch (up to ~70% SWE-bench)
- 45% fewer unnecessary refusals than Claude 3.5 Sonnet
- Launched with the Claude Code agentic coding tool
Best for
- Agentic coding and software-engineering tasks via Claude Code
- Math and science problems benefiting from extended thinking
- Front-end and full-stack web development
- Multimodal reasoning over text and images
- Tasks needing controllable reasoning depth and cost
How to access
| Provider | Model ID |
|---|---|
| Anthropic ↗ | claude-3-7-sonnet-20250219 |
| Amazon Bedrock ↗ | anthropic.claude-3-7-sonnet-20250219-v1:0 |
Claude Sonnet — every version
The full lineage of the Claude Sonnet line, newest first. Every version has its own page — click any to compare specs, benchmarks and pricing.
| Version | Released | Context | License |
|---|---|---|---|
| Claude Sonnet 4.6current | 2026-02-17 | 1M | Proprietary |
| Claude Sonnet 4.5 | 2025-09-29 | 1M | Proprietary |
| Claude Sonnet 4 | 2025-05-22 | 200K | Proprietary |
| Claude 3.7 Sonnet | 2025-02-24 | 200K | Proprietary |
| Claude 3.5 Sonnet (new) | 2024-10-22 | 200K | Proprietary |
| Claude 3.5 Sonnet | 2024-06-20 | 200K | Proprietary |
| Claude 3 Sonnet | 2024-03-04 | 200K | Proprietary |
FAQ
What made Claude 3.7 Sonnet different?
It was Anthropic's first hybrid reasoning model: a single model that can answer quickly or switch on extended thinking, with a configurable thinking budget up to 128K tokens.
How did it score on SWE-bench Verified?
62.3% in standard mode and up to 70.3% with extended thinking and additional scaffolding, leading agentic coding at launch.
What is the context window and max output?
200K token context window with up to 128K output tokens (which include extended-thinking tokens).
Is Claude 3.7 Sonnet still available?
No. The claude-3-7-sonnet-20250219 snapshot was retired on February 19, 2026. Anthropic recommends Claude Sonnet 4.6 as the drop-in replacement.