Claude receives 1 million tokens of support through the API to compete with Gemini 2.5 Pro for improved SEO.

Claude Sonnet 4 has received a significant upgrade, now capable of remembering up to 1 million tokens of context when accessed via API. This enhancement represents a fivefold increase from the previous limit, allowing Claude to retain over 75,000 lines of code or hundreds of documents in a single session. Previously, users had to submit information in smaller chunks, which often led to Claude losing context as it reached its limit. With the new 1 million token capacity, developers can create more sophisticated applications, as Claude can now remember more code than ever before. It is important to note that this context limit applies exclusively to Sonnet 4, while Opus 4.1 continues to operate under the old constraints due to its higher costs.

The rollout of the new context limit is currently underway via the Anthropic API for customers with Tier 4 and custom rate limits, with broader access expected in the coming weeks. Long context capabilities are also available in Amazon Bedrock and will soon be introduced to Google Cloud’s Vertex AI. With the ability to handle 1 million tokens, users can load entire codebases with all dependencies, analyse numerous documents simultaneously, and develop agents that maintain context across multiple tool calls. While pricing will adjust for prompts exceeding 200,000 tokens, prompt caching can help mitigate costs and reduce latency. Claude’s mobile and web applications are anticipated to receive the 1 million token context limit in the future. 

Categories: AI Model Upgrades, Context Limitations, Application Development 

Tags: Claude, Sonnet 4, Tokens, Context, API, Code, Documents, Anthropic, Amazon Bedrock, Google Cloud