What changed: Claude introduced a beta 'message threads' feature on the Messages API for long conversations and agent loops, in beta with the message-threads-2026-08-12 beta header. The API keeps the conversation on the server, so each later request sends only the new messages, lowering time to first token. A thread expires 5 minutes after its last request and is deleted no more than 1 hour after its last use.
Who is affected
Developers and businesses using the Messages API via Claude API, Claude Platform on AWS, Claude Models on Google Cloud (Anthropic managed), or Claude in Microsoft Foundry.
What it means for your business
For businesses using the Claude API, this means long-running conversations or agent loops can send smaller requests and get faster first-token responses using server-side message threads.
What to do
If your Claude API integration runs long conversations or agent loops, test message threads by adding the message-threads-2026-08-12 beta header, and design around a thread expiring 5 minutes after its last request and being deleted no more than 1 hour after its last use.
When it takes effect
The source gives no separate effective date; the feature became available in beta as of November 2, 2026 with the message-threads-2026-08-12 beta header.
Source: https://platform.claude.com/docs/en/release-notes/overview#november-2-2026
