Plugins

Libre WebUI supports three types of plugins:

• Chat - AI language models (OpenAI, Anthropic, Groq, etc.)
• Image Generation - Create images from text (ComfyUI, Flux)
• Text-to-Speech - Convert text to audio (OpenAI TTS, ElevenLabs)

Chat Plugins

Connect to cloud AI providers alongside local Ollama models.

Supported Providers

Provider	Models	API Key Variable	Notes
OpenAI	GPT-4o, GPT-4, o1, o3, o4 (110+ models)	OPENAI_API_KEY	Also provides TTS
Anthropic	Claude Opus 4.5, Claude 4 Sonnet/Opus	ANTHROPIC_API_KEY	Best for reasoning
Google Gemini	Gemini 2.0/2.5 Flash/Pro (55+ models)	GEMINI_API_KEY	Includes Imagen
Groq	Llama 3.1, Gemma, Qwen3	GROQ_API_KEY	Fastest inference
Mistral	Large, Medium, Codestral (71+ models)	MISTRAL_API_KEY	EU-based
OpenRouter	300+ models from all providers	OPENROUTER_API_KEY	Pay-per-token

Setup

Option 1: Environment variables (recommended for self-hosting)

Add API keys to backend/.env:

OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-ant-...
GROQ_API_KEY=gsk_...
GEMINI_API_KEY=...
MISTRAL_API_KEY=...
OPENROUTER_API_KEY=sk-or-...

Option 2: Per-user API keys (multi-user deployments)

Users can add their own API keys in Settings → Plugins → Configure. Keys are encrypted and stored per-user.

Enable plugins in Settings → Plugins.

Image Generation

Generate images using ComfyUI with Flux models.

ComfyUI Plugin

{
  "id": "comfyui",
  "name": "ComfyUI Flux",
  "type": "image",
  "endpoint": "http://localhost:8189/prompt",
  "capabilities": {
    "image": {
      "model_map": ["flux1-dev", "flux1-schnell"],
      "config": {
        "sizes": ["512x512", "768x768", "1024x1024", "1920x1080"],
        "default_size": "1024x1024"
      }
    }
  }
}

Setup

1. Install ComfyUI
2. Add Flux models to ComfyUI
3. Update the endpoint in plugins/comfyui.json to your ComfyUI server
4. Enable in Settings → Plugins

Usage

Click the image icon in chat or use the Imagine page to generate images.

Text-to-Speech

Convert AI responses to spoken audio.

OpenAI TTS

Uses OpenAI's text-to-speech API.

OPENAI_API_KEY=sk-...

Voices: alloy, ash, coral, echo, fable, onyx, nova, sage, shimmer

ElevenLabs

High-quality multilingual voices with 7 TTS models.

ELEVENLABS_API_KEY=...

Models: eleven_multilingual_v2, eleven_turbo_v2_5, eleven_flash_v2_5, and more

Voices (16 available): Rachel, Domi, Bella, Antoni, Josh, Adam, Arnold, Sam, and more

Formats: MP3, PCM, ulaw (5000 character limit per request)

Usage

1. Click the speaker icon on any message to hear it spoken
2. Configure voice and model in Settings → Text-to-Speech
3. Select between OpenAI TTS or ElevenLabs as your provider

Plugin Configuration

Plugins are JSON files in the plugins/ directory.

Plugin Structure

{
  "id": "provider-name",
  "name": "Display Name",
  "type": "completion|image|tts",
  "endpoint": "https://api.example.com/v1/...",
  "auth": {
    "header": "Authorization",
    "prefix": "Bearer ",
    "key_env": "API_KEY_VAR"
  },
  "model_map": ["model-1", "model-2"],
  "capabilities": {}
}

Plugin Types

Type	Purpose	Example
completion	Chat/text generation	OpenAI, Anthropic
image	Image generation	ComfyUI
tts	Text-to-speech	OpenAI TTS, ElevenLabs

Managing Plugins

Via UI

Settings → Plugins → Plugin Manager

• Enable/disable plugins
• Upload new plugins
• Configure settings

Via API

# List plugins
GET /api/plugins

# Enable plugin
POST /api/plugins/activate/:id

# Disable plugin
POST /api/plugins/deactivate

Creating Custom Plugins

Chat Plugin Example

{
  "id": "custom-llm",
  "name": "Custom LLM",
  "type": "completion",
  "endpoint": "https://your-api.com/v1/chat/completions",
  "auth": {
    "header": "Authorization",
    "prefix": "Bearer ",
    "key_env": "CUSTOM_API_KEY"
  },
  "model_map": ["model-a", "model-b"]
}

The API must follow the OpenAI chat completions format.

TTS Plugin Example

{
  "id": "custom-tts",
  "name": "Custom TTS",
  "type": "tts",
  "endpoint": "https://your-api.com/v1/audio/speech",
  "auth": {
    "header": "Authorization",
    "prefix": "Bearer ",
    "key_env": "CUSTOM_TTS_KEY"
  },
  "capabilities": {
    "tts": {
      "voices": ["voice-1", "voice-2"],
      "default_voice": "voice-1",
      "formats": ["mp3", "wav"]
    }
  }
}

Troubleshooting

Plugin not working:

• Check API key is set in .env
• Verify plugin is enabled in Settings
• Check server logs for errors

Image generation fails:

• Verify ComfyUI is running
• Check endpoint URL is correct
• Ensure Flux models are installed

TTS not playing:

• Check API key has credits
• Verify audio format is supported
• Check browser allows audio playback