Anthropic has launched "Message Threads" as a public beta for its Messages API, designed to reduce latency in long conversations. This feature transmits only new messages and the previous response ID after the initial full conversation is sent, cutting median time to first token by approximately 35% for contexts between 500k and 1M tokens. While token usage and costs remain the same, the feature aims to improve the performance of long-running AI agents, and it is already enabled by default for Claude Code users. AI
IMPACT Reduces latency for long-context AI agents, potentially improving user experience and enabling more complex conversational applications.
RANK_REASON Product feature launch for an API.
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →