Anthropic has announced a major update to its Claude Sonnet 4 AI model, increasing its context window to an unprecedented one million tokens—a scale that allows prompts longer than most books. The enhancement, aimed squarely at developer-focused businesses, marks a fivefold increase over Claude’s previous capacity and is double that of ChatGPT-5’s current limit. In practical terms, users can now provide up to 75,000 lines of code or approximately 750,000 words in a single prompt.
The expanded capability positions Claude to handle highly complex, data-intensive workflows. According to Anthropic, this includes large-scale code analysis, enabling developers to load entire codebases for review, gain insight into architecture, detect dependencies, and receive targeted improvement suggestions. It also enables document synthesis across massive datasets, such as collections of legal agreements or research papers, while maintaining contextual continuity throughout. Another application lies in building sophisticated, context-aware agents that can track multi-step workflows, API documentation, and extended interaction histories—even spanning hundreds of tool calls.
Anthropic has established itself as a leading AI model provider for enterprise development teams, with its technology powering tools like Windsurf, Microsoft’s GitHub Copilot, and Anysphere’s Cursor. Many developers see Claude as their primary AI coding companion. However, competition is intensifying—particularly from OpenAI’s ChatGPT-5, which offers a compelling subscription model alongside strong coding performance.
Speaking to TechCrunch, Brad Abrams, Anthropic’s product lead for Claude, noted that the broader context window significantly boosts performance on complex, long-duration coding assignments, where the AI must work independently for extended periods. “Claude can recall every step it has taken in long-horizon activities when it has a broad context window,” Abrams said.
The update follows last week’s release of Claude Opus 4.1, Anthropic’s most advanced AI model to date, which further enhanced its coding capabilities. Industry-wide, larger context windows are increasingly seen as a differentiator—Google’s Gemini 2.5 Pro offers a two-million-token window, while Meta’s Llama 4 Scout claims ten million tokens.
To offset the higher computing costs associated with processing such large inputs, Anthropic will introduce revised pricing for Claude Sonnet 4 API users. When prompts exceed 200,000 tokens, the input token price will double from $3 to $6 per million tokens, and the output token price will rise from $15 to $22.50 per million tokens.
