[FEATURE]:Add Support for Realtime API with WebSocket from OpenAI

[thetanav]

Feature hasn't been suggested before.

• I have verified this feature I'm about to request hasn't been suggested before.

Describe the enhancement you want to request

Add support to openai's Realtime API with WebSocket for existing models and remove older rest approach. This will make the responses ~30x faster for openai provider models.

Suggested edits:

• use the ai-sdk-openai-websocket-fetch package from ai sdk as drop in replacemen.
• extra add a onFinish in streamText to close the ws

For side by side comparision:

• github/vercel-labs

open ai docs
ai sdk docs

[github-actions]

This issue might be a duplicate of existing issues. Please check:

• [FEATURE]: Add support to openai ws apis #14887: Same feature request by the same author — "Add support to openai ws apis" (now closed)

[thetanav]

This is the demo showing difference in time https://ai-sdk-openai-websocket-demo.vercel.app/

[thetanav]

@rekram1-node should i work on this issue?

[nick-potts]

its not really drop in because codex uses its own fetch impl

[itpropro]

Can't wait for this to drop in opencode!

[maurits150]

I see noticeable higher performance on websockets (when trying out the pi harness) on codex. This feature is pretty relevant.