A simple playground for testing and comparing different language models using the Fastify framework and fastify-lm plugin.

LM Playground is a Node.js application that allows you to:

• Test multiple language model providers (OpenAI, Google, Claude, Deepseek, Llama, Mistral)
• Compare responses from different models to the same prompt
• Access results via a simple HTTP endpoint

• Support for multiple LLM providers
• Easy configuration via environment variables
• Simple REST API to retrieve model responses
• Development mode with hot reloading

• Node.js
• pnpm (Package Manager)

1. Clone the repository
2. Install dependencies:

pnpm install

3. Create a .env file with your API keys:

OPENAI_KEY=your_openai_key
CLAUDE_KEY=your_claude_key
DEEPSEEK_KEY=your_deepseek_key
LLAMA_KEY=your_llama_key
MISTRAL_KEY=your_mistral_key

Start the development server:

pnpm dev

The server will run at http://localhost:3333

You can test the API with curl:

pnpm curl

You can modify the following in src/play.ts:

• Available models and providers
• System prompt
• Test question

• fastify - Web framework
• fastify-lm - Plugin for language model integration
• dotenv - Environment variable management
• pino - Logging
• TypeScript - Type safety

ISC