Human Versions Sonnet 4.6 | TechCrunch

Anthropic has released a new version of its mid-range Sonnet model, to keep up with the company’s four-month refresh cycle. In a post announcing the new modelAnthropic emphasized improvements in programming, following instructions, and computer use.

Sonnet 4.6 will be the default model for Free and Pro plan users.

The Sonnet 4.6 beta will include a contextual window with 1 million icons, twice the size of the largest window previously available for Sonnet. Anthropic described the new context window as “enough to accommodate entire codebases, long contracts, or dozens of research papers in a single request.”

The release comes after only two weeks Opus 4.6 launchand an updated Haiku model will likely follow in the coming weeks.

The launch comes with a new set of benchmark scores, including OS World for PC Usage and SWE-Bench for Software Engineering. But perhaps most impressive is his score of 60.4% on the ARC-AGI-2, which aims to measure skills specific to human intelligence. The score puts the Sonnet 4.6 above most comparable models, though it still lags behind models like the Opus 4.6, Gemini 3 Deep Think, and one improved version of GPT 5.2.