Grok Think Fast API and MCP: real-time voice conversations

Grok Think Fast is a real-time voice model: it receives and returns audio as a stream, so a conversation runs without pauses for recognition and synthesis. On NeuralSpace it is available through the /v1/realtime WebSocket endpoint, and the key is issued on the API keys page.

What the model can do

The model holds a live voice conversation: it hears speech, answers with a voice and keeps a natural pace. It suits voice assistants, call centers and apps that need a conversation without delays. You pay per minute of audio sent and received.

Endpoint

Connect over WebSocket: wss://neuralspace.pro/v1/realtime?model=grok-voice-think-fast-2.0&lang=en. The language is set with the lang parameter (ru, en, ar, zh). Authorization uses an nsk-… API key.

MCP tools

The voice stream itself runs over WebSocket, while the MCP server at POST /mcp exposes the utility tools list_models (model and price list) and get_balance (balance check). This lets an agent check the rate and funds before starting a voice session.

How to connect

Sign up on NeuralSpace, create a key on the API keys page and pass it when opening the WebSocket connection. Full documentation is on the /v1/docs page.

FAQ

Is it REST or WebSocket?

WebSocket: audio streams in both directions.

Which languages are supported?

ru, en, ar and zh — the lang parameter in the connection address.

Where do I get an API key?

In the dashboard on the API keys page.