Anthropic launches message threads in public beta, letting Messages API apps send only new messages
Anthropic introduced message threads in public beta on November 2, 2026. After an initial full-conversation request, Messages API applications on the Claude Platform can send only new messages. The feature targets latency-sensitive sessions such as long-running coding agents and computer-use agents.
Anthropic introduced message threads in public beta on November 2, 2026. Messages API applications on the Claude Platform can now send only new messages after an initial full-conversation request, according to a summary published by Unite.AI.
What changed
Under the new feature, an application sends the full conversation once. After that, requests within the same session carry only the new messages.
Target use cases
Anthropic describes message threads as intended for active sessions where latency matters. The use cases named are:
- Long-running coding agents
- Computer-use agents, whose screenshots remain with the thread
- Document- and image-heavy workflows
Anthropic also reports that message threads are on by default for most Claude Code users. The summary is cut off at that point, so further details on that claim are not available here.
Not a new model
Message threads is an API feature, not a new model. No context window, pricing, benchmark or model version details accompany the announcement.
What this means
Message threads change how applications send conversation state to the Messages API. Instead of resending the full conversation on each request, an app sends it once and then sends only new messages. Anthropic positions this for sessions where latency matters, including agents that accumulate screenshots, documents or images over time. Because the feature is in public beta, developers building such workflows on the Claude Platform can test it now.
Related Articles
Anthropic Study: 50 Analysts Using Claude Code Followed Agent's Hidden Preference 86% of the Time
Anthropic published a randomized controlled trial on October 27, 2026 in which 50 professional analysts using Claude Code followed their agent's secretly assigned preference 86% of the time on open analytical judgment calls. The study asks whether experts make the consequential decisions in AI-assisted work or defer to their agents.
Anthropic adds two-sentence rule to Opus 5.5 system prompt to stop Claude cloning existing games
Anthropic confirmed it added two sentences to the Claude Opus 5.5 system prompt instructing Claude not to recreate a specific game owned by someone else. Claude Code users have since reported refusals on original projects in GitHub issues.
Users report Claude refusing game recreation and reverse-engineering tasks; Anthropic silent
Multiple users claim Anthropic's Claude Code and/or Opus 5.5 now refuse to recreate specific existing games and some reverse-engineering tasks, including use of the Reverse Engineer Anything (REA) tool. Anthropic has made no official statement, and the rule text circulating online is unconfirmed.
Anthropic, OpenAI, Cloudflare and Google each release AI agents that hunt security bugs in code
Anthropic, OpenAI, Cloudflare and Google have each released an AI agent that hunts security bugs in source code. MarkTechPost reports that all four use the same find, disprove, then patch loop, and its comparison also covers 12 more tools.
Comments
Loading...