Claude message threads enter beta: long conversations send only new messages, cutting time to first token
Synopsis
Anthropic released message threads in beta on the Messages API, keeping long conversations and agent loops on the server so each later request sends only the new messages, which shrinks requests and lowers time to first token. Previously such multi-turn interactions had to resend the full history each time. A thread expires 5 minutes after its last request and is deleted no more than 1 hour after its last use, and it is available on the Claude API, Claude Platform on AWS, Claude Models on Google Cloud (Anthropic managed), and Claude in Microsoft Foundry.
Interpretation
Long conversations and agent loops on the Messages API can send only new messages, making requests smaller and lowering time to first token. Previously each request in a multi-turn conversation carried the full history; a thread keeps that history on the server. The capability is in beta behind the message-threads-2026-08-12 beta header and is available on the Claude API, Claude Platform on AWS, Claude Models on Google Cloud (Anthropic managed), and Claude in Microsoft Foundry.
The server holds the conversation, and each later request submits only the new messages. Message threads are the mechanism carrying this behavior, moving conversation state from the client request body to the server. A thread expires 5 minutes after its last request and is deleted no more than 1 hour after its last use.
Perspective
The demonstrated scope is long conversations and agent loops on the Messages API, enabled with a beta header, on the Claude API, Claude Platform on AWS, Claude Models on Google Cloud (Anthropic managed), and Claude in Microsoft Foundry.
The release notes give no specific figures for request-size reduction or time-to-first-token improvement, and a thread expiring 5 minutes after its last request and being deleted no more than 1 hour after its last use constrains sessions with long gaps.