Anthropic introduced message threads for the Messages API on November 2, 2026, according to its release notes.reported This beta feature targets long conversations and agent loops by reducing request sizes and lowering the time to the first token.Anthropic lets the server remember chat history so apps send less data.Anthropic implements stateful server-side context retention to minimize payload overhead in agent loops.
Evidence: reported
Claude API Release Notes page updated: Long conversations and agent loops on the Messages API can now send smaller requests and gpagewatch-anthropic
The claim
Anthropic states that the API now retains conversation history on the server side. This architecture allows clients to transmit only new messages in subsequent requests rather than resending the full context. The company indicates this approach yields a lower time to the first token.
The response
The release notes define strict lifecycle rules for these threads. A thread expires five minutes after its last request. The system deletes the thread no later than one hour after its final use. No third-party commentary or counterpoints appear in the source material.
The practical upshot
Product teams can access this feature via the `message-threads-2026-08-12` beta header.reported It is available on the first-party Claude API, Claude Platform on AWS, Claude Models on Google Cloud, and Claude in Microsoft Foundry.
Evidence: reported
Claude API Release Notes page updated: Long conversations and agent loops on the Messages API can now send smaller requests and gpagewatch-anthropic
Claims versus evidence
Anthropic claims that message threads reduce request size and improve time to first token. The evidence is the architectural shift to server-side storage, which eliminates the need to resend historical messages. The source does not provide benchmark data or independent verification of these performance gains. Limits include the five-minute expiration window and the one-hour deletion cap. These constraints suggest the feature is designed for short-lived or intermittent agent interactions rather than persistent, long-running sessions. The beta status implies the implementation may change.
Why it matters
- Product teams building agent loops can reduce payload sizes by leveraging server-side context retention.
- The five-minute expiration rule requires developers to redesign session management for long-running tasks.
- Availability across AWS, Google Cloud, and Microsoft Foundry allows consistent testing across multi-cloud deployments.
- The beta header `message-threads-2026-08-12` signals that the interface may change before general availability.
Topics: Anthropic, Claude API, agent loops, message threads, beta features
Updated
Then / now / next
Clients had to resend full conversation histories for each API request.
The API stores conversations server-side, allowing smaller, incremental requests in beta.
Watch for the removal of the beta header and any changes to expiration policies.
Story timeline
- Then Clients had to resend full conversation histories for each API request.
- 01 Nov 2026 DeepSeek open sources Super-Agent as a research prototype, a proactive coding agent it says can monitor group chats for coding issues and suggest solutions (Debby Wu/Bloomberg)
- 02 Nov 2026 Anthropic's internal AI use, at retail prices, now exceeds its employee pay
- 02 Nov 2026 DeepSeek releases a "creator mode" that allows DeepSeek Harness users to quickly create custom plugins, like a 3D model renderer and a video editing workbench (Debby Wu/Bloomberg)
- 02 Nov 2026 Claude API Release Notes page updated: Long conversations and agent loops on the Messages API can now send smaller requests and g
- 02 Nov 2026 Goldman Sachs pulls AI coding tools from traders and bankers
- 03 Nov 2026 v2.1.317
- Now The API stores conversations server-side, allowing smaller, incremental requests in beta.
- Next Watch for the removal of the beta header and any changes to expiration policies.
The strongest doubt
Counterpoint from the sources
The source material does not provide independent benchmarks to verify the claimed reduction in time to first token.
Ask the article
Local keyword search of this article; no generated answers.Did this article give you the evidence you needed?
Your choice stays on this device. No public tally.Terms in this story
Time to first token
The latency between sending a request and receiving the first generated token.
Agent loops
Automated sequences where an AI model iteratively calls tools or APIs.
Beta header
An API parameter used to opt into experimental features.
Questions readers ask
How long does a message thread persist?
A thread expires five minutes after the last request and is deleted within one hour of final use.
Which platforms support message threads in beta?
The feature is available on the Claude API, AWS, Google Cloud, and Microsoft Foundry.
Is this feature generally available?
No, it is currently in beta and requires a specific beta header.
Connected stories
Reading receipt
- Message threads are available in beta on the Claude API and major cloud providers as of November 2, 2026.
- The feature keeps conversations on the server, so subsequent requests send only new messages.
- Threads expire five minutes after the last request and are deleted within one hour of final use.
Still open: How does the feature behave if a client fails to send a request within the five-minute window?
Still open: Will the expiration and deletion policies change before general availability?