Anthropic has expanded the capacity of its mid-tier large language model, Claude Sonnet 4, to handle a one-million-token context window—five times its previous limit. The feature, released Tuesday in public beta on the Anthropic API and Amazon Bedrock, lets developers submit more than 75,000 lines of code or hundreds of documents in a single request, eliminating the need to split large projects into smaller segments. The company is keeping existing prices for prompts of 200,000 tokens or fewer at $3 per million input tokens and $15 per million output tokens, while charging $6 and $22.50 respectively for larger requests. Anthropic says early enterprise users are applying the extended context to full-repository code reviews, large-scale document synthesis and multi-step autonomous agents. The upgrade intensifies competition with OpenAI and Google, whose flagship models already offer million-token windows. Anthropic argues that Claude’s retrieval accuracy—rated at 100% on internal “needle-in-a-haystack” tests—offsets higher usage costs. Broader rollout to additional customer tiers and cloud platforms is expected in the coming weeks.