Skip to main content

AI Providers

DotAgents integrates with multiple AI providers for chat, speech-to-text, and text-to-speech. Configure one or more providers to power your agents.


Chat Providers

These providers power the LLM that drives your agents' reasoning and responses.

OpenAI

SettingValue
API Keysk-...
Base URLhttps://api.openai.com/v1 (default)
ModelsGPT-4o, GPT-4o-mini, o1, o3, GPT-4 Turbo

Best for: General-purpose, highest quality reasoning.

Groq

SettingValue
API Keygsk_...
Base URLhttps://api.groq.com/openai/v1 (default)
ModelsLlama 3.3 70B, Llama 3.1 8B, Mixtral 8x7B

Best for: Fast inference, real-time voice interactions.

Google Gemini

SettingValue
API KeyGoogle AI API key
ModelsGemini 2.0 Flash, Gemini Pro, Gemini Ultra

Best for: Multimodal tasks, large context windows.

Local Models

Use any OpenAI-compatible local model server:

ServerBase URL
Ollamahttp://localhost:11434/v1
LM Studiohttp://localhost:1234/v1
llama.cpphttp://localhost:8080/v1

Best for: Privacy-first, offline use, no API costs.

Speech-to-Text (STT) Providers

OpenAI Whisper

  • High accuracy across 30+ languages
  • Requires API key and internet connection
  • Model: whisper-1

Groq Whisper

  • Same Whisper model, accelerated by Groq hardware
  • Significantly faster than OpenAI's endpoint
  • Requires Groq API key

Parakeet (Local)

  • Runs locally using ONNX runtime
  • No API key or internet required
  • Good accuracy for English
  • Model runs on CPU

Text-to-Speech (TTS) Providers

OpenAI TTS

  • 6 voices: Alloy, Echo, Fable, Onyx, Nova, Shimmer
  • High quality, natural-sounding
  • Models: tts-1, tts-1-hd

Groq (Orpheus)

  • Fast inference via Groq hardware
  • Multiple voice options
  • Low latency for real-time playback

Google Gemini TTS

  • Multiple voices
  • Natural prosody
  • Supports multiple languages

Kitten TTS

  • Custom voice engine
  • Configurable via Kitten API

Supertonic TTS

  • Custom voice engine
  • Additional voice options

Configuration

Via the UI

  1. Go to Settings > Providers
  2. Select your provider
  3. Enter your API key
  4. Choose a model from the dropdown
  5. (Optional) Set a custom base URL

Via Configuration Files

In .agents/models.json:

{
"providers": {
"openai": {
"apiKey": "sk-...",
"baseUrl": "https://api.openai.com/v1"
},
"groq": {
"apiKey": "gsk_..."
},
"gemini": {
"apiKey": "AI..."
}
},
"defaultModel": {
"provider": "openai",
"model": "gpt-4o"
}
}

Per-Agent Model Override

Each agent can use a different model/provider:

{
"modelConfig": {
"provider": "groq",
"model": "llama-3.3-70b-versatile"
}
}

Model Presets

DotAgents includes pre-configured model bundles that set the provider, model, and recommended settings together. Select a preset in Settings > Models for quick setup.

Rate Limiting

DotAgents handles API rate limits gracefully:

  • Automatic retry with exponential backoff
  • Rate limit status shown in the UI
  • Per-provider rate limit tracking

Next Steps