Anthropic says message threads is in public beta, reducing repeated context resend work and cutting median TTFT about 35% for 500k to 1M tokens.
Anthropic has made message threads available in public beta on the Claude Platform, a change aimed at reducing latency in long-running agent sessions by avoiding the need to resend full conversation history on every turn.
According to Anthropic’s announcement published on 2 November 2026, message threads let agents send only what is new since the last turn, rather than retransmitting the full conversation and tool history each time. The company said that in sessions reaching hundreds of thousands of tokens, repeatedly sending earlier messages can add time before Claude begins responding. It said requests with 500,000 to 1 million tokens see median time to first token, or TTFT, fall by about 35% with message threads enabled by default for most Claude Code users.
The company said developers can use the feature by adding the thread parameter to Messages API requests. The first request includes the full conversation and starts a thread, while each later request includes only the new messages and the ID of the previous response. Anthropic said threads are held in memory between turns, encrypted and not written to disk, and that each thread expires five minutes after its last request, at which point the client must resend the full conversation to start a new one.
Anthropic also said the feature does not reduce token usage, because Claude still reads the full conversation on every turn.
Source link







