Anthropic has announced the release of Claude Sonnet 4.6, its latest AI model, on February 17, 2026.

The model features a 1 million-token context window in beta and is designed to improve coding, reasoning, and computer use capabilities.

It is now the default in claude.ai and Claude Cowork for free and pro users, with pricing unchanged from the previous version.

Early user feedback indicates significant improvements in instruction following and task consistency, often surpassing earlier models like Claude Opus 4.5.

The company reports extensive safety evaluations showing Sonnet 4.6 to be as safe or safer than recent models, with a focus on prosocial behavior and reduced risks of high-stakes errors.

Performance benchmarks, such as OSWorld, demonstrate notable progress in AI’s ability to interact with software and perform complex tasks like navigating spreadsheets and web forms, though the model still lags behind human experts.

The company emphasizes ongoing safety and performance enhancements across its platforms.

🔗 Source: Anthropic

Recent Anthropic developments