Anthropic’s Claude Sonnet 4.6 launches with faster speeds and 1M token context

Anthropic has launched Claude Sonnet 4.6 and set it as the default model for free and Pro users on claude.ai and Claude Cowork. The company calls it its most capable Sonnet model so far and says it includes a 1 million token context window in beta. Anthropic frames Sonnet 4.6 as a practical daily driver and notes it can be faster than the premium Opus 4.6 in many cases.

Anthropic says developers with early access preferred Sonnet 4.6 over prior Sonnet releases and even Opus 4.5. The company claims benchmark results show it outperforming Google Gemini 3 Pro and OpenAI GPT 5.2 on agentic financial analysis and office tasks, and also beating Opus 4.6 in those categories. Anthropic also highlights improved coding performance aimed at a broader set of users.

It is available via web, API, and major cloud platforms at the same pricing as the previous model. Free usage limits reset every five hours, while the Pro plan costs $20 per month or $17 per month when billed annually.