Anthropic speeds up Claude Code with message threads. But your token bill stays the same.

Anthropic has unveiled a new feature called message threads aimed at speeding up lengthy Claude Code sessions. This feature reduces the amount of conversation data repeatedly sent to the API, enabling developers to receive quicker responses as their coding sessions expand. The beta version of message threads is now accessible in Anthropic's Messages API and is automatically enabled for most Claude Code users.

The feature works by temporarily storing conversation history on Anthropic's servers, allowing subsequent requests to include only new messages and a reference to the previous response. This approach cuts down on latency, particularly during extended coding sessions, where the conversation history tends to grow. However, it's important to note that Claude still processes the entire conversation on each turn, so token usage remains unchanged.

While message threads doesn't inherently reduce API costs, developers can combine it with Claude's existing prompt caching feature to further minimize token usage and processing costs. The new feature is particularly beneficial for long-running coding agents, computer-use agents that repeatedly capture screenshots, and applications handling large volumes of documents or images. By avoiding repeated uploads, message threads can significantly reduce the amount of data transmitted during a session.

To implement the feature, developers must use a new thread parameter in the Messages API, indicating whether to create a new thread or continue an existing one. The Python SDK also offers a simpler approach through a thread helper, which automatically manages conversation history and API parameters.