Claude stops resending full history in long chats: message threads keep the conversation server-side so later requests carry only new messages
Synopsis
Anthropic has 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 the release notes describe as smaller requests and a lower time to first token. Previously such multi-turn calls had to resend the full history each time. The feature uses the message-threads-2026-08-12 beta header, a thread expires 5 minutes after its last request and is deleted no more than 1 hour after 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 no longer require the client to resend the whole history each turn; a later request carries only the new messages. Previously such multi-turn requests carried the full conversation every time, so request size grew with each turn. The capability is offered in beta with the message-threads-2026-08-12 beta header, and the release notes state the result as smaller requests and a lower time to first token.
The mechanism keeps the conversation on the server, with the API holding the dialogue state while the client appends new messages. Message threads are that server-side session mechanism, replacing client-side replay of the entire history. A thread expires 5 minutes after its last request and is deleted no more than 1 hour after its last use, and the feature covers the Claude API, Claude Platform on AWS, Claude Models on Google Cloud (Anthropic managed) and Claude in Microsoft Foundry.
Perspective
The feature is in beta and requires the message-threads-2026-08-12 beta header in requests, and it applies to the Claude API, Claude Platform on AWS, Claude Models on Google Cloud (Anthropic managed) and Claude in Microsoft Foundry. The intended reader is an API user building long conversations and agent loops.
The benefit is described only as smaller requests and a lower time to first token, with no figures or comparison baseline; a thread expires 5 minutes after its last request and is deleted no more than 1 hour after last use, a limit that constrains sessions with long gaps.