Sign in
← AI PulseJul 23, 2026

Wire · news · Multi-source brief

xAI Introduces Voice Agent API and Multi-Agent Research Capabilities

xAI has launched a Voice Agent API for real-time voice conversations and introduced multi-agent research capabilities with the **grok-4.20-multi-agent** model.

By Illumora Editorial · Jul 23, 2026

Synthesized from multiple allowlisted primaries on the same event. Lanes →

Brief drafted by Illumora’s editorial model from the linked primary source. Ops desk reviews flagged pieces. How we write →

Read the source →xAI Docs — Voice Agent API | SpaceXAI Docs
Save

xAI has released a Voice Agent API designed to facilitate real-time voice conversations over WebSocket. This API is billed based on audio duration and text input messages. Concurrently, xAI has introduced multi-agent research functionality, enabling the orchestration of multiple AI agents for complex research tasks.

Key Points

  • The Voice Agent API supports real-time voice conversations via WebSocket.
  • Billing for the Voice Agent API includes a charge of $0.05 per minute for audio and $0.004 per text input message.
  • The Voice Agent API operates in the us-east-1 region.
  • The multi-agent research feature allows Grok to coordinate AI agents for deep, multi-step research.
  • To utilize multi-agent research, users specify grok-4.20-multi-agent as the model name in API requests.
  • The multi-agent research feature is currently in beta, and its API interface and behavior may change.

Context

According to xAI documentation, the Voice Agent API charges for every conversation.item.create event sent from the client, with exceptions for response.create events and function_call_output items. Items with input_audio or audio content are billed by the audio meter. The multi-agent capability, as detailed by xAI, allows specialized agents to collaborate on tasks such as searching the web, analyzing data, and synthesizing findings.

Why It Matters

The introduction of the Voice Agent API provides developers with tools for integrating real-time voice interactions into applications, while the multi-agent research capability offers a method for handling complex, multi-step research tasks through coordinated AI agents. These features expand the range of applications for xAI's models.

What To Do

  • Review the pricing structure for the Voice Agent API, noting the $0.05 per minute audio charge and $0.004 per message text input fee.
  • Explore the Voice Agent API documentation for details on WebSocket integration and function calling support.
  • Test the multi-agent research feature by specifying grok-4.20-multi-agent in API requests for research-oriented tasks.
  • Note that the multi-agent research feature is in beta and its API may undergo changes.

Keep Exploring

/atlas/grok-family