Anthropic has released Claude Sonnet 4.6, an upgraded version of its Sonnet model, boasting enhanced capabilities in coding, computer use, long-context reasoning, agent planning, knowledge work, and design.

The new model features a 1 million token context window in beta and is now the default for users on the Free and Pro plans in claude.ai and Claude Cowork, with pricing remaining consistent with Sonnet 4.5.

Developers have shown a strong preference for Sonnet 4.6 over its predecessor, citing improvements in consistency and instruction following, with many even preferring it over the previous top-tier model, Claude Opus 4.5.

Sonnet 4.6 demonstrates significant advancements in computer use, performing comparably to human-level capabilities in tasks like navigating complex spreadsheets and filling out multi-step web forms, as measured by the OSWorld benchmark.

The model also shows improved resistance to prompt injection attacks, a critical safety consideration for AI that interacts with computer systems.

In terms of coding, Sonnet 4.6 is preferred by users for its better context reading and consolidation of shared logic, leading to a less frustrating development experience.

The expanded 1 million token context window allows Sonnet 4.6 to process entire codebases or extensive documents, enabling more effective long-horizon planning, as evidenced by its strategic performance in the Vending-Bench Arena simulation.

Early customers have reported more polished visual outputs, better layouts, and improved design sensibility in frontend code and financial analysis generated by Sonnet 4.6, requiring fewer iterations to achieve production-quality results.

Databricks, Replit, Cursor, GitHub, Cognition, Windsurf, Hebbia, Box, Pace, Bolt, Rakuten, Zapier, Convey, Triple Whale, Harvey, and other partners have noted significant improvements in their respective applications using Sonnet 4.6, highlighting its enhanced reasoning, coding, and computer use capabilities.

On the Claude Developer Platform, Sonnet 4.6 supports adaptive and extended thinking, along with context compaction in beta. The API now features automatic code execution for filtering web search results, improving efficiency and response quality.

While Sonnet 4.6 offers a powerful and cost-effective alternative, Anthropic still recommends Opus 4.6 for tasks requiring the deepest reasoning, such as complex codebase refactoring and multi-agent coordination.

👉 Read the full article on Anthropic