openai OpenAI News ·

WebSockets speed up OpenAI agent workflows in Responses API

ai
feature

OpenAI has integrated WebSockets into the Responses API to accelerate agentic workflows by reducing API overhead and improving model latency. This deep dive into the Codex agent loop demonstrates the performance gains achieved through this enhancement. The update is relevant for developers building complex AI-driven applications that require fast, iterative responses.

Enhancements (1)
  • WebSockets reduce API overhead and latency for agentic workflows

    WebSockets have been integrated into the Responses API, significantly reducing API overhead and improving model latency. This optimization is demonstrated through a deep dive into the Codex agent loop, showcasing how connection-scoped caching further enhances performance.

Read the original announcement →

https://openai.com/index/speeding-up-agentic-workflows-with-websockets

Related releases