--- title: "Claude Message Threads for Your Company | By The People Tech" description: "Message threads, in public beta on the Messages API, let agents send only what is new each turn. Anthropic reports median time to first token down about 35%." url: https://bythepeople.tech/claude/whats-new/claude-message-threads --- [Claude for Business](/claude) [What’s New](/claude/whats-new) Message threads in the Claude Messages API New feature Released 2 November 2026 · 4 min read # Long agent sessions start answering _faster_. Anthropic has released message threads in public beta on the Messages API. An agent sends the full conversation once, then only what is new on each turn. Anthropic reports that median time to first token drops by about 35% on Claude Code requests of 500k to 1M tokens. At a glance Status Public beta Time to first token About 35% lower (median) Token usage Unchanged Rates and limits Standard rates apply Thread expiry 5 min after last request On this page * [What changed](#changed) * [What it means](#means) * [What to do this week](#steps) * [Questions](#questions) Published by [Fatjon (Tony) Kalemaj](/about-us/fatjon-tony-kalemaj) COO & Head of Applied AI 01 · What changed ## What Anthropic shipped. What changed: until now, every request in an agent loop resent every earlier message and tool result. Once a conversation reached hundreds of thousands of tokens, that resending added time before Claude started responding. With message threads, the API keeps the conversation on the server between turns, so each later request is smaller. Claude still reads the full conversation on every turn, so token usage stays the same: the gain is speed, and the bill is unchanged. ### 1. Send only what is new 1\. The first request carries the full conversation and starts a thread. Each later request carries only the new messages and the ID of the previous response. ### 2. Faster first token 2\. Anthropic reports that requests with 500k to 1M tokens see median time to first token drop by about 35%, with threads on by default for most Claude Code users. ### 3. Same token usage 3\. Claude still reads the full conversation on every turn, so token usage stays the same. Standard Claude token rates and existing rate limits apply. ### 4. Held in memory, encrypted 4\. Anthropic says each thread is held in memory between turns, encrypted and not written to disk. A thread is deleted no more than 1 hour after its last use. ### 5. Five-minute expiry 5\. A thread expires 5 minutes after its last request. The client then resends the full conversation to start a new one. ### 6. Opt in per request 6\. Add the thread parameter to Messages API requests and use the message-threads-2026-08-12 beta header. ### 7. Four places to run it 7\. Available on the Claude API, Claude Platform on AWS, Claude in Microsoft Foundry and Claude Models on Google Cloud (Anthropic managed). 02 · What it means ## What it means for _your_ seat. If your firm builds agents on the Claude API that work through long material, such as a data room review or a stack of credit agreements, message threads can shorten the wait before each answer. It does not lower the bill, because token usage stays the same. Anthropic says it helps most where an agent carries a lot of context and latency matters. ### Leadership Read this as a speed improvement for long agent sessions. Costs do not fall: standard token rates and existing rate limits still apply. Ask your engineers which internal agents carry hundreds of thousands of tokens of context, since those are where Anthropic says the gain is largest. ### IT and engineering Add the thread parameter and the message-threads-2026-08-12 beta header. The first request sends the full conversation; later requests send only new messages plus the previous response ID. Build the fallback: after 5 minutes with no request the thread expires and your client must resend the full conversation. ### Risk and compliance The API now keeps the conversation on the server between turns. Anthropic says threads are held in memory, encrypted, not written to disk, expire 5 minutes after the last request and are deleted no more than 1 hour after last use. Check that against your data retention policy and vendor terms before production use, and note the feature is in beta. ### Your teams Anthropic says message threads is on by default for most Claude Code users, so engineers in long coding sessions have nothing to switch on. Anyone using an internal agent built on the API will only see a change once your engineering team adopts the feature. 03 · This week ## What to do this week. 0 of 3 done 1. 01 · List agents that carry long context : Find internal agents whose sessions grow large, such as a data room review loop or an IC memo drafting agent with many documents. Anthropic says these gain the most. 2. 02 · Send the data handling terms to compliance : Share the stated terms: held in memory, encrypted, not written to disk, deleted no more than 1 hour after last use. Get a decision before any production traffic. 3. 03 · Trial threads on one agent in test : Add the thread parameter and beta header to one agent, handle the 5-minute expiry, and compare time to first token in your own logs before and after. Sources Facts and dates are Anthropic’s own. What it means and what to do are our view, from the firms we work with. Drafted with Claude from these sources, checked against them figure by figure, and published 2 November 2026. * [Anthropic: Make long-running agents faster with message threads](https://claude.com/resources/articles/message-threads) * [Claude Platform release notes](https://platform.claude.com/docs/en/release-notes/overview) ## The short answers. ### What are Claude message threads? Message threads is a public beta feature of the Claude Messages API. The API keeps the conversation on the server, so after the first request an agent sends only the new messages and the ID of the previous response, which reduces time to first token on long sessions. ### Do Claude message threads reduce token costs? No. Claude still reads the full conversation on every turn, so token usage stays the same, and standard Claude token rates and existing rate limits apply. The benefit Anthropic describes is lower latency on long sessions. ### How long does a Claude message thread last? A thread expires 5 minutes after its last request, and the client then resends the full conversation to start a new one. Anthropic says threads are held in memory, encrypted and not written to disk, and are deleted no more than 1 hour after their last use. ### Where can I use Claude message threads? Message threads is available in public beta on the Claude API, Claude Platform on AWS, Claude in Microsoft Foundry and Claude Models on Google Cloud (Anthropic managed). Requests use the thread parameter and the message-threads-2026-08-12 beta header. Read next [30 October 2026 : Claude Opus 4.8 fast mode deprecation](/claude/whats-new/claude-opus-4-8-fast-mode-deprecation) [28 October 2026 : Smart reports, surveys and pooled group budgets](/claude/whats-new/claude-enterprise-smart-reports-pooled-budgets) [23 October 2026 : Claude chat and Cowork unify for Enterprise](/claude/whats-new/claude-cowork-chat-unified-enterprise) Newsletter ## Keep up with _every_ release. One email when something ships or retires that matters to a company: what changed, what it means, the date and what to do. Unsubscribe any time. Work email How did you hear about us? How did you hear about us? ChatGPT Claude Perplexity Another AI assistant Google or Bing search LinkedIn YouTube Someone referred me An event or webinar Our newsletter Somewhere else ## Want this working inside your firm? We run Claude workshops on your own documents and stay on as your teams adopt each release. [Book 30 minutes with Tim](https://calendly.com/timminer-1/chat-with-tim) [hello@bythepeople.tech](mailto:hello@bythepeople.tech?subject=Claude%3A%20Message%20threads%20in%20the%20Claude%20Messages%20API)