DeepGemini provides a flexible, OpenAI-compatible API for orchestrating multiple AI models, targeting developers and researchers needing to integrate diverse AI capabilities into their applications. It simplifies complex AI workflows by allowing users to chain models, define AI roles, and manage multi-model discussions through a unified interface.

How It Works

The core of DeepGemini is its multi-model orchestration engine, built on FastAPI. It supports a pluggable architecture for various AI providers, including DeepSeek, Claude, Gemini, Grok3, OpenAI, OneAPI, and OpenRouter. Users can define "Relay Chains" to sequence models for reasoning and execution steps, and create AI "roles" with specific personalities and skills. These roles can then form "discussion groups" to engage in various modes like brainstorming, debate, or SWOT analysis, enabling sophisticated AI agent interactions.

Quick Start & Requirements

• Installation: git clone https://github.com/sligter/DeepGemini.git && cd DeepGemini && uv sync
• Prerequisites: Python 3.11, FastAPI, SQLAlchemy, Alembic, aiohttp. Requires API keys for supported AI providers.
• Configuration: Copy .env.example to .env and set ALLOW_API_KEY and ALLOW_ORIGINS.
• Run: uv run uvicorn app.main:app --host 0.0.0.0 --port 8000
• Docker: Recommended via docker-compose up -d.
• Docs: Web management UI available at /dashboard.

Highlighted Details

• OpenAI API compatibility for seamless integration.
• Supports multiple AI providers and custom integrations.
• Advanced features include multi-role discussions and customizable workflows.
• Built-in web management UI for configuration and monitoring.
• Real-time streaming responses.

Maintenance & Community

The project is actively maintained by sligter. Support and questions are handled via GitHub Issues.

Licensing & Compatibility

Licensed under the MIT License, permitting commercial use and integration with closed-source applications.

Limitations & Caveats

The project is primarily designed for Linux/Mac environments for direct Docker usage, with specific instructions for Windows PowerShell. While it supports multiple AI providers, the quality and availability of specific models depend on the respective provider's API.