AI SDK
@aisdk
Use OpenAI WebSockets with the AI SDK to save up to 40% time to first byte. https://t.co/7gNm7ut1YK
48retweets
View on XAI SDK Adds OpenAI WebSocket Transport for 40% Faster Streaming
OpenAI· Updated
Vercel's AI SDK added OpenAI WebSocket transport that cuts time to first byte by up to 40%. Swap in a WebSocket fetch adapter and streaming responses arrive faster by skipping repeated connection handshakes.
AI SDK, Vercel's open-source toolkit for building AI applications, added WebSocket transport for OpenAI connections. Instead of opening a new HTTP connection for each API call, WebSockets maintain a persistent connection that eliminates TCP and TLS handshake overhead on every request - cutting time to first byte by up to 40%.
The gain compounds for apps that make frequent OpenAI calls — chat interfaces and agent loops pay TCP and TLS handshake costs on every HTTP request, and those stack up across a conversation. A persistent WebSocket connection absorbs that overhead once, so the latency benefit grows with call frequency.
Drop the WebSocket transport into an existing AI SDK project — streamText, message handling, and response formatting stay unchanged, and the persistent connection delivers the latency improvement.
Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →
