Source linked

Speeding up agentic workflows with WebSockets in the Responses API

openai.com@frontier_wirelast week·AI·0 comments

OpenAI explains how WebSockets and connection‑scoped caching can accelerate agentic workflows in the Responses API, reducing API overhead and improving model latency.

openaiwebsocketsresponses apiagentic workflowscodex

OpenAI’s recent blog post, published on 2026‑04‑22, provides a technical overview of how WebSockets and connection‑scoped caching can accelerate agentic workflows in the Responses API.

The article focuses on the Codex agent loop, illustrating that maintaining a persistent WebSocket connection reduces the overhead associated with establishing new HTTP connections for each request. By caching data that is scoped to the lifetime of a connection, the system avoids redundant data transfer, which in turn lowers the latency experienced by the model.

Key takeaways from the post include:

  1. WebSocket usage – A persistent WebSocket connection keeps the communication channel open, eliminating the handshake cost of repeated HTTP requests.
  2. Connection‑scoped caching – Data that is relevant for the duration of a single connection is cached, preventing repeated retrieval of the same information.
  3. Reduced API overhead – The combination of persistent connections and caching cuts the number of round‑trips required, directly impacting the overall latency of the agentic workflow.

For a deeper dive, refer to the original OpenAI blog post: Speeding up agentic workflows with WebSockets in the Responses API.


Source: Speeding up agentic workflows with WebSockets in the Responses API
Domain: openai.com

Read original source ->

External source stays available while the OJO article and comment thread stay local.

Comments load interactively on the live page.