Anthropic has released a new version of its mid-size Sonnet model, maintaining the company’s four-month update cycle. The announcement highlighted significant improvements in coding, instruction-following, and computer use capabilities. Sonnet 4.6 will serve as the default model for users on both the Free and Pro plans.
The beta release features a dramatically expanded context window of one million tokens. This is twice the size of the largest window previously available for the Sonnet model. Anthropic described this new capacity as sufficient to hold entire codebases, lengthy contracts, or dozens of research papers within a single request.
This release follows just two weeks after the launch of Opus 4.6, with an updated Haiku model expected to arrive in the coming weeks. The launch is accompanied by a new set of record benchmark scores. Notable achievements include high marks on OS World for computer use and SWE-Bench for software engineering.
Perhaps the most impressive result is a 60.4% score on the ARC-AGI-2 benchmark, which is designed to measure skills specific to human intelligence. This score places Sonnet 4.6 above most comparable models, though it still trails behind others like Opus 4.6, Gemini 3 Deep Think, and one refined version of GPT 5.2.