- Anthropic released Message Threads as a public beta on the Claude platform, cutting time to first token by about 35 percent for conversations between 500,000 and one million tokens.
- The feature transmits only what has changed since the last message instead of resending the entire conversation history on every API request.
- Coding agents, computer-use agents, and workflows that process many documents or images benefit the most, according to Anthropic.
- Threads stay encrypted in memory and expire after five minutes of inactivity; total token usage is unchanged since Claude still reads the full history.
What Happened
Anthropic released Message Threads as a public beta on its Claude platform, a change to how the API handles long conversations (The Decoder, Nov. 3, 2026). Previously, every API request had to resend the entire conversation history; Message Threads instead transmits only what changed since the last message.
Why It Matters
Anthropic says the change cuts time to first token by about 35 percent for requests between 500,000 and one million tokens, addressing a latency problem long-running agent sessions create. The feature builds on Anthropic’s push toward persistent, context-heavy agent tools such as Claude Managed Agents.
Technical Details
Coding agents, computer-use agents, and workflows that process large volumes of documents or images see the biggest gains, according to Anthropic. A thread stays encrypted in memory between requests and expires after five minutes of inactivity. Token usage itself doesn’t change, since Claude still reads the full conversation history each time; only the transmission step is reduced.
Who’s Affected
Developers building long-running coding and computer-use agents on Claude, including Claude Code users, stand to see the most latency improvement. The feature is available on the Claude platform and through AWS, Microsoft Foundry, and Google Cloud, extending it to enterprise customers on those platforms.
What’s Next
Message Threads remains in public beta, with full technical details published in Anthropic’s developer documentation. Anthropic has not stated when the feature will move to general availability.
