Anthropic’s message threads cut the wait on the longest Claude Code sessions by about 35 percent
Release notes dated 2 November put the Messages API feature in beta. A coding agent sends the conversation once, then only the new turn. Token prices do not change. A thread expires after five minutes of silence and is not eligible for zero data retention or HIPAA.
Short2 minEasy
Priya Shah2 min read
- Anthropic
- Claude Code
- Agents
- API

Photo: AnthropicSource page
Anthropic on Monday put message threads into public beta on the Claude Platform, so a long-running agent can stop re-uploading its whole conversation on every turn. The platform release notes list the change under 2 November 2026. The product page says a coding agent may make hundreds of requests in one session, and that each request normally resends every earlier message and tool result. Once that history reaches hundreds of thousands of tokens, the resend adds time before Claude starts answering. With threads, the first request carries the full conversation and starts the thread. Later requests send only what is new, plus the ID of the previous response. Anthropic says that with threads on by default for most Claude Code users, requests of 500,000 to 1 million tokens see median time to first token fall by about 35 percent.
The saving is latency, not the bill. The product page says Claude still reads the full conversation on every turn, so token usage stays the same, and standard rates and existing rate limits apply. Between turns the thread is held in memory, encrypted, and not written to disk. It expires five minutes after the last request, and the client then has to send the full conversation again. The docs add that the API deletes a thread no more than one hour after its last use, and that each stored response restarts the five minutes. If a continue request can no longer find the thread, it returns a 404, and the client starts over. The docs say threads are not eligible for zero data retention, and that organizations with HIPAA readiness enabled cannot use them. They are in beta on every Claude model, on the Claude API, on Claude Platform on AWS, on Claude Models on Google Cloud where Anthropic manages the service, and in Microsoft Foundry. The beta header named in the release notes is message-threads-2026-08-12.
The product page points the feature at three kinds of session: a coding agent whose tool results and file edits keep lengthening the transcript, a computer-use agent whose screenshots would otherwise be uploaded again on every turn, and document or image work where the request itself is large. The docs say there is no API to list, read, or delete a thread, and that the application remains the source of truth for the conversation. The page does not say what share of Claude Code users are in the “most” for whom threads are already on.


