Anthropic’s Claude Sonnet 4 model has been upgraded to support a 1 million token context window, marking a significant enhancement over its previous 200,000 token limit. This substantial increase, equivalent to about 750,000 words or more than 75,000 lines of code, allows for processing entire codebases, multiple research papers, and book-length content in a single request. This development unlocks new capabilities for large-scale code analysis, document synthesis, and sophisticated context-aware AI agents.
Available in public beta through Anthropic’s API and Amazon Bedrock, with Google Cloud Vertex AI support anticipated soon, the extended context window facilitates the analysis of complete software projects and comprehensive document collections without losing context. As a result, developers can manage complex workflows more efficiently, maintaining state across large tasks without needing to break them into smaller parts, thereby reshaping traditional development workflows.
However, this expanded capacity comes with increased pricing for prompts exceeding 200,000 tokens, doubling costs to $6 per million input tokens, which is higher than competitors like Google’s Gemini model. While the price adjustment affects larger context executions, Anthropic offers strategies like prompt caching and batch processing to mitigate costs.
Claude’s enhancement not only keeps it competitive against models like OpenAI’s GPT-5 but also presents new challenges regarding security and compliance. Managing large-scale input could potentially expose sensitive data, underscoring the importance of robust security protocols.
Overall, the upgrade solidifies Claude Sonnet 4’s position in the AI landscape by enabling enterprise-level processing and analysis capabilities, thus catering to the demands of AI coding platforms and enterprise customers seeking comprehensive solutions for processing vast information efficiently.

Collections

Claude Sonnet 4 now supports a 1 million token context window, up from 200,000 tokens, enabling processing of entire codebases and book-length content in single requests. Available through Anthropic's API and Amazon Bedrock, this upgrade allows for large-scale code analysis and document synthesis without losing context. However, costs double to $6 per million input tokens for prompts exceeding 200,000 tokens, though strategies like prompt caching can help manage expenses. The enhancement positions Claude competitively against models like GPT-5 while raising new security considerations for enterprise use.

Anthropic Expands Claude Sonnet 4 with 1 Million Token Context Window