DeepClaude is an open-source project that provides an OpenAI-compatible API for advanced AI model chaining, specifically combining DeepSeek R1 with models like Claude 3.7 Sonnet or Gemini 2.5 Pro. It targets developers and users looking to leverage state-of-the-art AI for code generation and content creation with enhanced performance and cost-efficiency, offering a unified interface for various LLM providers.

How It Works

The project utilizes FastAPI to create an OpenAI-compatible backend. It chains two models: a "reasoning" model (primarily DeepSeek R1) and a "target" model (like Claude 3.7 Sonnet or Gemini 2.5 Pro). This architecture is inspired by research suggesting that using DeepSeek R1 as an architect and a more capable model like Claude 3.5 Sonnet as an editor achieves state-of-the-art results on benchmarks at a significantly lower cost. It supports both streaming and non-streaming outputs and allows configuration of various parameters like temperature.

Quick Start & Requirements

• Install: Clone the repository and install dependencies using uv.
• Run: Use uvicorn app.main:app --port 8000 for local or server deployment.
• Prerequisites: API keys for DeepSeek (recommended via Volcengine), Claude (Anthropic), or Gemini (Google AI Studio). Python environment managed by uv.
• Setup: Access the configuration interface at http://127.0.0.1:8000/config after running.
• Docs: Detailed setup and configuration instructions are available in the README.

Highlighted Details

• Offers distinct model combinations: deepclaude (DeepSeek R1 + Claude 3.7 Sonnet) for programming, deepgeminipro (DeepSeek R1 + Gemini 2.5 Pro) and deepgeminiflash (DeepSeek R1 + Gemini 2.0 Flash) for content creation (free options available).
• Supports OpenAI-compatible API endpoints, allowing integration with various chat clients like Cherry Studio, NextChat, ChatBox, and LobeChat.
• Features a graphical configuration interface (version 1.0+) for easier setup, replacing .env files.
• Includes GitHub Actions for automatic fork synchronization and Docker image building.

Maintenance & Community

The project is actively maintained with frequent updates addressing model compatibility, configuration, and performance. Community interaction is encouraged via the provided email and website.

Licensing & Compatibility

The project is licensed under the MIT License, permitting commercial use and integration with closed-source applications.

Limitations & Caveats

The project notes that token consumption statistics for non-streaming requests were not fully implemented at the time of the last update. Some DeepSeek R1 providers might require specific prompts or configuration adjustments (e.g., disabling "native reasoning support") if the reasoning content is not correctly parsed.