Anthropic Launches Message Threads to Cut Claude Response Times

Anthropic has announced the launch of a new feature called Message Threads for its artificial intelligence platform, Claude. The tool is currently available as part of a Public Beta and aims to speed up response times during lengthy conversations and complex tasks that require processing large volumes of data. This update specifically targets developers who use Claude tools to build AI agents, helping to reduce the time needed to begin generating answers, particularly during long programming sessions or when handling documents, images, and multi-step tasks.

How Message Threads Function

The feature operates by changing how conversation history is sent to AI models. Instead of resending the entire chat log with every new request, developers can now send only the most recent message along with the identifier of the model’s previous response. This mechanism allows the system to access the cached context of the prior conversation without needing to transmit all older messages across the network again, thereby reducing the time spent on request processing.

Performance Improvements for Developers

According to Anthropic, using Message Threads can reduce the Time-to-First-Token latency by approximately 35% when dealing with large requests ranging between 500,000 and one million tokens. This improvement is significant for developers relying on AI agents for complex operations, such as writing code over extended periods, controlling computers, and analyzing large documents and images.

Data Retention and Storage Rules

Anthropic confirmed that this feature does not alter API usage prices or limits, as the Claude model still processes the full conversation context when preparing an answer. Conversations are temporarily stored in server memory, and the message thread expires five minutes after the last request. If this period elapses, developers must resend the conversation history. The company stated that these data points are not saved on traditional storage units and are completely deleted within one hour of the last use, in accordance with the feature’s published policy.

Supported Platforms and Restrictions

Message Threads are currently available in beta via the Claude Platform. Support extends to Amazon Web Services, Microsoft Foundry, and Anthropic-managed models on Google Cloud. The company noted that the feature is enabled by default for most Claude Code users. However, it is not available for accounts that rely on Zero Data Retention policies or those requiring compliance with HIPAA standards for health data.