Anthropic launches message threads for Claude, cuts time-to-first-token roughly 35%
Anthropic just rolled out a new feature called "Message Threads" for its Claude AI, now in public beta.
Instead of sending the whole chat history every time, developers can now just send the latest message and previous response ID, making long conversations much smoother.
Anthropic says Message Threads cuts Claude's time-to-first-token by roughly 35% for massive requests containing 500,000 to 1 million tokens, which is a big deal for things like coding marathons or working through complex documents.
Message threads stores chats up-to-five-minutes
Message Threads temporarily stores chats for up to five minutes after your last message (and deletes them within an hour for security).
You'll find this feature on platforms like Claude, AWS, Microsoft Foundry, and Anthropic-managed models on Google Cloud.
For most users of Claude Code, it's already switched on by default, but if your account has strict privacy rules (like HIPAA), you won't have access just yet.
The best part: all these speed boosts come with no extra cost or changes to usage limits.