AI RESEARCH
Speeding up agentic workflows with WebSockets in the Responses API
OpenAI Blog
•
A deep dive into the Codex agent loop, showing how WebSockets and connection-scoped caching reduced API overhead and improved model latency.