Claude Sonnet 4.6
Claude Sonnet 4.6 becomes the default Free and Pro model, preferred over Opus 4.5 by 59% of users, at unchanged $3/$15 pricing.
Full upgrade across coding, computer use, long-context reasoning, agent planning and design, with a 1M-token window (beta). Developers preferred it to Sonnet 4.5 about 70% of the time and to Opus 4.5 59% of the time. SWE-bench Verified 80.2% (with prompt modification); ARC-AGI-2 60.4% at high effort.
- Date
- Tuesday, 17 February 2026
- Lab
- Anthropic
- Kind
- model
- Access
- closed API
- Price
- $3/$15 per M tokens in/out, unchanged from Sonnet 4.5
Figures
| Measure | Value | Measured by |
|---|---|---|
| SWE-bench Verified | 80.2% with prompt modification | company |
| ARC-AGI-2 | 60.4% high effort | company |
| Claude Code preference vs Opus 4.5 | 59% 70% preferred it over Sonnet 4.5 | company |
Free tier gained file creation, connectors, skills and compaction. Anthropic's Sonnet 5 page later restates Sonnet 4.6 at 34.6% (no tools) and 46.8% (with tools) on Humanity's Last Exam. OSWorld-Verified figure on the 4.6 page was a chart and is not listed.
Sources
This record was checked against its sources on 6 October 2026. How we check