Anthropic highlights Claude Sonnet’s Opus-class reasoning ability and performance 4.6

Anthropic has improved the programming and use capabilities of the computer with the model Claude Sonnet 4.6, whose performance stands out in tasks that require deeper reasoning.

Claude Sonnet 4.6 is a complete update to Anthropic’s AI model, billed as the “most capable” of the Sonnet models to date, with a performance in line with the Opus category.

This model offers a leap in programming as a result of improvements in consistency and following instructionsbut its main advance stands out in the autonomous use of a computeras Anthropic has indicated in a statement.

“Early users of Sonnet 4.6 are experiencing human-level capability in tasks like navigating a complex spreadsheet or filling out a multi-step web form, before integrating it all into multiple browser tabs,” he says.

Offers a million token context window –at least in beta– “enough” space to “host entire codebases, large contracts, or dozens of research papers in a single request.”

Reasoning abilities are another point that Anthropic has highlighted. Sonnet 4.6 “reasons effectively across that context,” improving long-term planning.

Furthermore, it supports both adaptive thinking like him extended thinkingas well as the context compaction in beta, a feature that automatically summarizes older context as conversations approach limits.

The company assures that “Opus 4.6 remains the strongest choice for tasks that require deeper thinkingsuch as refactoring the code base, coordinating multiple agents in a workflow and problems where it is essential to get it right,” he added.