Anthropic Message Threads Cuts Repeat Context Sending in Messages API
A new public beta lets clients send only new messages per turn, reporting a 35 percent drop in median time-to-first-token for very long requests.
Anthropic introduced Message Threads, a public beta for its Messages API, on 2 November 2026. The first request sends the full conversation with a thread creation parameter; subsequent requests carry only the new message plus a previous_message_id. Context is held temporarily in encrypted server-side memory, expires five minutes after the last request, and stale thread IDs return a 404 so clients must resend full context.
Anthropic reports the median time-to-first-token falling about 35 percent for requests between 500,000 and 1 million tokens, with median request size shrinking from roughly 800 KB to 1.8 KB. Token usage and pricing are unchanged, and the feature is active by default for most Claude Code users. Anthropic notes it is not available to zero-data-retention and HIPAA customers.