llm-discovery

LLM model discovery and tracking system for real-time monitoring of available models across multiple providers (OpenAI, Google AI Studio/Vertex AI, Anthropic).

Features

• Zero-Configuration Quick Start: Use prebuilt model data without API keys (max 24h old)
• Real-time Model Discovery: Fetch available models from multiple LLM providers
• Automatic Data Source Switching: Seamlessly switches between API and prebuilt data
• Multi-format Export: Export model data in JSON, CSV, YAML, Markdown, and TOML formats
• Change Detection: Track model additions and removals over time
• Data Source Transparency: See whether data comes from API or prebuilt sources with timestamps
• CI/CD Integration: Easy integration with GitHub Actions and other CI/CD systems
• Python API: Use as a library in your Python applications
• Offline Mode: Cache-first operation for offline usage

Installation

Method 1: uvx (No Installation Required - Recommended)

# Fetch and cache models
uvx llm-discovery update

# Display cached models
uvx llm-discovery list

Method 2: pip

pip install llm-discovery

Quick Start

Without API Keys (Zero Configuration)

Try llm-discovery instantly without any API keys using prebuilt model data (updated daily):

# Display models using prebuilt data (no API keys required)
uvx llm-discovery list

# Export prebuilt data
uvx llm-discovery export --format json --output models.json

The data source and timestamp are displayed in the output, showing whether data comes from API or prebuilt sources.

With API Keys (Real-time Data)

Set up API keys for the providers you want to use:

# OpenAI
export OPENAI_API_KEY="sk-..."

# Google AI Studio
export GOOGLE_API_KEY="AIza..."

# Google Vertex AI (alternative to AI Studio)
export GOOGLE_GENAI_USE_VERTEXAI=true
export GOOGLE_APPLICATION_CREDENTIALS="/path/to/gcp-credentials.json"

Basic Usage

# Fetch and cache models from all providers
uvx llm-discovery update

# Display cached models in a table
uvx llm-discovery list

# Detect changes from previous snapshot
uvx llm-discovery update --detect-changes

# Export to JSON
uvx llm-discovery export --format json --output models.json

Python API

import asyncio
from llm_discovery import DiscoveryClient
from llm_discovery.models.config import Config

async def main():
    # Load configuration from environment variables
    config = Config.from_env()
    client = DiscoveryClient(config)

    # Fetch models from all providers
    provider_snapshots = await client.fetch_all_models()

    # Display all models
    for provider in provider_snapshots:
        for model in provider.models:
            print(f"{model.provider_name}/{model.model_id}: {model.model_name}")

asyncio.run(main())

Documentation

For detailed documentation, visit our documentation site or see:

• Quick Start Guide
• Installation Guide
• API Reference (Technical Contract)
• CLI Reference (Technical Contract)

Development

Setup

# Clone the repository
git clone https://github.com/drillan/llm-discovery.git
cd llm-discovery

# Install with development dependencies
uv sync --all-extras --all-groups

# Run tests
uv run pytest

# Run linting
ruff check .

# Run type checking
mypy llm_discovery/

License

MIT License - see LICENSE for details.

Contributing

Contributions are welcome! Please see CONTRIBUTING.md for guidelines.