Back to docsRoadmap

Realtime voice

Nimbus will bridge WebSocket connections directly to the OpenAI Realtime API, giving you sub‑second voice responses inside your existing agent flows.

What's planned

  • •Persistent WebSocket sessions with automatic reconnect
  • •Server‑side VAD so you never send silent audio chunks
  • •Function‑calling passthrough — voice triggers your existing tools
  • •Interruption handling so the agent stops speaking when you cut in

Status

The WebSocket bridge is in active development. We are targeting a private alpha in Q3 2026. If you want early access, reach out through the dashboard support channel.

This feature requires an OpenAI API key with Realtime API access. Pricing follows OpenAI's per‑minute audio token rates.