Claude Sonnet 4.6: Enhanced Coding, Computer Use, and a 1M Token Context Window

Introducing Claude Sonnet 4.6

Feb 17, 2026

Anthropic has just released Claude Sonnet 4.6, the latest and most advanced iteration of their Sonnet series. This new model brings significant upgrades across multiple fronts, including coding, computer use, long-context reasoning, agent planning, knowledge work, and design. Notably, it also features a 1M token context window in beta, which is a game-changer for handling large documents and complex tasks.

Key Upgrades

• Coding Skills: Sonnet 4.6 demonstrates much-improved coding abilities, with enhancements in consistency, instruction following, and overall performance. Early adopters have shown a strong preference for Sonnet 4.6 over its predecessor, Sonnet 4.5, and even over the more advanced Claude Opus 4.5 from November 2025.
• Computer Use: The model excels in using software that lacks modern APIs, thanks to its ability to interact with a computer as a human would-clicking and typing. This capability is crucial for automating tasks in legacy systems.
• Context Window: A 1M token context window (in beta) allows the model to handle longer documents and maintain context over extended interactions, making it ideal for knowledge work and long-form content creation.

Performance Highlights

Sonnet 4.6's performance improvements are not just incremental; they represent a significant leap in practical utility. Here’s what you can expect:

• Consistency and Accuracy: The model is more consistent in following instructions and generating accurate code, reducing the need for manual corrections.
• Real-World Tasks: It performs tasks that previously required an Opus-class model, including economically valuable office tasks (as evaluated by ArtificialAnalysis.ai).
• Safety Evaluations: Extensive safety evaluations have shown Sonnet 4.6 to be as safe as or safer than previous models. Safety researchers noted that it has a "broadly warm, honest, prosocial, and at times funny character" with strong safety behaviors.

Computer Use: A Breakthrough in Automation

One of the most exciting aspects of Claude Sonnet 4.6 is its enhanced computer use capabilities. This feature addresses a common challenge for organizations that rely on legacy software-software that lacks modern APIs and is difficult to automate.

• OSWorld Benchmark: Our models have made steady gains on the OSWorld benchmark, which evaluates AI performance across real software (Chrome, LibreOffice, VS Code, etc.) running in a simulated environment. The model interacts with these applications using a virtual mouse and keyboard, much like a human user.
• Practical Applications: Early users of Sonnet 4.6 have reported significant improvements in automating tasks that were previously cumbersome or error-prone. This capability can streamline workflows and reduce the need for custom connectors.

Availability and Pricing

Claude Sonnet 4.6 is now the default model on both the Free and Pro plans at claude.ai and in Claude Cowork. The pricing remains the same as Sonnet 4.5, starting at $3 per million tokens for the Free plan and $15 per million tokens for the Pro plan.

Conclusion

Claude Sonnet 4.6 represents a significant step forward in AI capabilities, particularly in coding and computer use. Its enhanced performance, combined with a larger context window, makes it a powerful tool for developers and organizations looking to automate complex tasks. Whether you're writing code, managing legacy systems, or handling large documents, Sonnet 4.6 is worth exploring.