Lemonade Server

Overview

Lemonade Server provides optimized local LLM inference with support for GPU and NPU hardware acceleration. It offers an OpenAI-compatible API that seamlessly integrates with Continue and other open-source platforms.

Installation

Download and install Lemonade Server from lemonade-server.ai.

Configuration

Lemonade Server is available directly in the Continue UI as a provider. You can select it from the model provider dropdown without manual configuration.

Option 1: Using the Continue UI (Recommended)

1. Click on the model selector dropdown in Continue
2. Select "Add Model"
3. Choose "Lemonade Server" from the provider list
4. Continue will automatically configure the connection

Option 2: Manual Configuration

If you need custom settings, you can manually configure Lemonade:

models:
  - name: Lemonade
    provider: lemonade
    model: <MODEL_NAME>
    apiBase: http://localhost:8000/api/v1/
```
</Tab>
<Tab title="JSON (Deprecated)">
```json title="config.json"
{
  "models": [
    {
      "title": "Lemonade",
      "provider": "lemonade",
      "model": "<MODEL_NAME>",
      "apiBase": "http://localhost:8000/api/v1/"
    }
  ]
}
```
</Tab>

Getting Started

1. Install Lemonade Server: Download from lemonade-server.ai
2. Start the server: Launch Lemonade Server (runs on http://localhost:8000/api/v1/ by default)
3. Add to Continue: Select Lemonade Server from the model provider dropdown in Continue
4. Load a model: Choose your preferred model through the interface

Hardware Support

Lemonade Server automatically detects and optimizes for available hardware:

• NPU: Neural Processing Unit acceleration for efficient inference
• GPU: Full GPU acceleration support
• CPU: Optimized CPU fallback when accelerators are unavailable

Key Features

• OpenAI-compatible API for seamless integration
• Support for popular model formats
• Automatic hardware detection and optimization
• Integration with Continue, Open WebUI, Gaia, and AnythingLLM
• Active open-source community

View the source