Anthropic launched message threads in beta on the Claude Messages API to cut time to first token

Archive entry: this covers an event from November 2, 2026, and we published it on November 8, 2026. The details come from sources dated at the time and may have changed since.

On November 2, 2026, Anthropic launched message threads in beta on the Claude Messages API, as described in Anthropic's article on message threads. The API keeps the conversation on the server, so each later request sends only the messages added since the previous response, plus the ID of that response. Anthropic said this lowers time to first token in long agent loops.

What changed

• Message threads were in beta and used the message-threads-2026-08-12 beta header.
• A thread expired 5 minutes after its last request and was deleted no more than 1 hour after its last use.
• Between turns, each thread was held in memory, encrypted and not written to disk.
• Claude still read the full conversation on every turn, so token usage stayed the same.

What it meant for builders

Agent builders with very long contexts could get noticeably faster responses by sending only the added turns, at the same token cost. Anthropic said that requests of 500k to 1M tokens saw median time to first token drop by about 35%, and that message threads were on by default for most Claude Code users.

Availability

Message threads were available on the Claude API, Claude Platform on AWS, Claude Models on Google Cloud (Anthropic managed) and Claude in Microsoft Foundry. Anthropic's Claude Platform release notes were a second source for the launch.

Sources

• Make long-running agents faster with message threads (Anthropic (Claude blog), November 2, 2026, official)
• Claude Platform release notes - Claude Platform Docs (Anthropic (Claude Platform Docs), November 2, 2026, official)

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.