Overview
- Claude Sonnet 4.6 is live as the default across claude.ai, Claude Cowork and Claude Code, the API and major cloud partners, with the free tier gaining file creation, connectors and context compaction.
- A beta 1 million‑token context window lets the model handle entire codebases, lengthy contracts or dozens of research papers within a single session.
- Anthropic reports developers preferred Sonnet 4.6 over Sonnet 4.5 about 70% of the time and over Opus 4.5 in roughly 59–60% of tests, citing better instruction following and fewer hallucinations.
- Computer‑use capabilities advanced to near‑human performance on OSWorld‑style tasks such as multi‑step web forms and spreadsheet navigation, though expert users still outperform the model.
- Sonnet pricing remains $3 per million input tokens and $15 per million output tokens, while Opus 4.6 continues as the recommended choice for the deepest reasoning and highest‑precision work.