Skip to content

AI Models

Configure language models, speech models, embeddings, and batch processing for your AI agents.

Overview

The AI Models section allows you to configure all AI models used across your workspace. Add custom API keys, adjust model parameters, and manage access to different model providers.

Model Categories

  • Language Models

    Configure GPT-4, Claude, Gemini, and other LLMs for conversational AI.

    Configure LLMs

  • :material-batch-processing:{ .lg .middle } Batch Models

    Set up batch processing for high-volume async AI tasks.

    Batch Processing

  • ASR Models

    Configure Automatic Speech Recognition (speech-to-text) models.

    ASR Setup

  • TTS Models

    Set up Text-to-Speech voices for AI agent responses.

    TTS Voices

  • :material-phone-voice:{ .lg .middle } Realtime Voice Models

    Configure real-time voice conversation models for phone/SIP.

    Realtime Voice

  • Embeddings

    Manage embedding models for RAG and semantic search.

    Embeddings

Supported Providers

Choose a provider from the creation chooser for the model category. A provider tile and the model names offered by that provider are different selections; an existing configuration may use a legacy provider no longer offered for new configurations.

  • Language Models — current provider chooser and model parameters.
  • ASR Models — speech recognition providers.
  • TTS Models — speech synthesis providers.
  • Realtime Voice — realtime model types and transport configuration.
  • Embeddings — Google AI, OpenAI and Azure OpenAI embedding configurations.

Configure the required credential before selecting it in a model form.