News
Anthropic Releases Claude Sonnet 4.6 with Record Benchmarks and Million-Token Context
Anthropic's latest mid-tier model, Sonnet 4.6, debuts with record scores in software engineering and computer use benchmarks, plus a doubled context window of one million tokens. The release becomes the new default for free and pro users.
Key Takeaways
- Sonnet 4.6 becomes the default model for free and pro Claude users with record SWE-Bench and OS World scores
- Context window doubles to 1 million tokens in beta — enough for entire codebases or dozens of research papers
- ARC-AGI-2 score of 60.4% outperforms most competing models, trailing only Opus 4.6, Gemini 3 Deep Think, and GPT 5.2
- Release follows Opus 4.6 by two weeks, with an updated Haiku model expected in coming weeks
DE
DT Editorial AI··via techcrunch.com