Anthropic has released Claude Sonnet 4.6, featuring improved coding skills, computer use abilities, and a 1M token context window, with performance approaching Opus-level intelligence at a lower price point.
Anthropic has unveiled Claude Sonnet 4.6, positioning it as their most capable Sonnet model to date with significant improvements across coding, computer use, long-context reasoning, agent planning, knowledge work, and design capabilities.
Enhanced Coding and Computer Use Performance
The new model brings substantial upgrades to coding functionality. Early access developers have shown a strong preference for Sonnet 4.6 over its predecessor, with many even favoring it to Claude Opus 4.5, Anthropic's November 2025 frontier model. Users report that Sonnet 4.6 reads context more effectively before modifying code, consolidates shared logic rather than duplicating it, and demonstrates less overengineering and "laziness" compared to earlier models.
Computer use capabilities have seen particularly notable advancement. Sonnet 4.6 shows major improvements in navigating complex spreadsheets, filling out multi-step web forms, and managing tasks across multiple browser tabs. The model achieves human-level capability in these areas, though it still lags behind the most skilled humans at computer use tasks.
Benchmark Performance and Context Window
Sonnet 4.6 features a 1M token context window in beta, enabling it to handle entire codebases, lengthy contracts, or dozens of research papers in a single request. The model demonstrates effective reasoning across this extended context, excelling at long-horizon planning tasks.
In the Vending-Bench Arena evaluation, which tests business management over time with competitive elements, Sonnet 4.6 developed a novel strategy of heavy initial investment followed by a sharp pivot to profitability in the final stretch. This approach helped it finish well ahead of competing models.
Pricing and Availability
Pricing remains consistent with Sonnet 4.5 at $3/$15 per million tokens. The model is now the default for Free and Pro plan users on claude.ai and Claude Cowork. Sonnet 4.6 is available across all Claude plans, Claude Cowork, Claude Code, the API, and major cloud platforms.
Safety and Risk Considerations
Anthropic conducted extensive safety evaluations, concluding that Sonnet 4.6 is "as safe as, or safer than" recent Claude models. The model exhibits "a broadly warm, honest, prosocial, and at times funny character" with strong safety behaviors and no major concerns around high-stakes misalignment.
However, computer use capabilities introduce new risks, particularly prompt injection attacks where malicious actors hide instructions on websites. Sonnet 4.6 shows improved resistance to these attacks compared to its predecessor, performing similarly to Opus 4.6 in safety evaluations.
Enterprise Applications
Early enterprise adopters report significant improvements across various use cases:
- Financial Services: 15 percentage point improvement in heavy reasoning Q&A
- Insurance: 94% accuracy on computer use benchmarks for workflows like submission intake
- iOS Development: Superior spec compliance and architecture with modern tooling adoption
- Frontend Development: Perfect design taste requiring less iteration for production-quality results
Developer Platform Updates
The Claude Developer Platform now supports adaptive thinking, extended thinking, and context compaction in beta for Sonnet 4.6. Context compaction automatically summarizes older context as conversations approach limits, effectively increasing usable context length.
API enhancements include automatic code generation for filtering and processing search results, improving response quality and token efficiency. Additional features like code execution, memory, and programmatic tool calling are now generally available.
Performance-to-Cost Ratio
Anthropic emphasizes the extraordinary performance-to-cost ratio of Sonnet 4.6, noting how rapidly Claude models have evolved in recent months. The model provides a viable alternative for teams currently using Opus, offering frontier-level reasoning in a more cost-effective form factor.
For the first time, Sonnet brings frontier-level reasoning capabilities to a smaller, more affordable model, making advanced AI accessible to a broader range of applications and users.
Related Announcements:
- Anthropic and the Government of Rwanda sign MOU for AI in health and education
- Anthropic and Infosys collaborate to build AI agents for telecommunications and other regulated industries
- Anthropic opens Bengaluru office and announces new partnerships across India
Comments
Please log in or register to join the discussion