Anthropic has introduced a new 'message threads' feature for its Claude API, designed to improve efficiency for agent loops and long conversations. This feature caches conversations on the server, allowing subsequent requests to send only new messages rather than the entire conversation history. While this can reduce the time to first token by up to 35% for very long sessions, the threads are ephemeral, expiring five minutes after the last request. Users are advised to maintain their own persistent copies of conversations, as the API's caching is described as 'best-effort'. The feature is available across various Claude models and platforms, including AWS and Google Cloud, and does not alter pricing or rate limits. AI
IMPACT Improves efficiency for AI agents and long conversations by reducing redundant data transmission.
RANK_REASON This is a feature update to an existing API, not a new model release or significant industry shift.
- Anthropic
- AWS
- Claude API
- Claude Code
- Claude Haiku-4-5
- Claude Platform
- Claude Sonnet 4.5
- Fable 5.1
- Google Cloud
- Messages API
- Microsoft Foundry
- Mythos 5.1
- Opus 5.5
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
