Speeding up agentic workflows with WebSockets in the Responses API
WebSockets just cut agent latency. Here's how OpenAI's Responses API is changing the speed math for production AI workflows.
Why it matters
OpenAI is shipping infrastructure optimizations (WebSockets + connection-scoped caching) that materially reduce API overhead for agentic systems. This is a developer-facing feature drop that affects how teams deploy agents in production, signaling OpenAI's focus on agent performance as a competitive differentiator.
The key facts
10 to knowWebSockets integration in Responses API
Connection-scoped caching reduces API overhead
Model latency improvements for agent loops
Published Apr 22 2026 (recent release)
Codex agent loop optimization focus
Feature: WebSocket support in Responses API
Use case: Codex agent loop optimization
Benefit: Reduced API overhead and improved model latency
Target: Agentic workflows at scale
Technology: Connection-scoped caching implementation
Go to the source
OpenAI Blogopenai.com
Publisher excerpt: A deep dive into the Codex agent loop, showing how WebSockets and connection-scoped caching reduced API overhead and improved model latency.