DeepClaude
By combining ai models, we can realize the use needs of using different models for programming, content creation, and daily experiments.
DeepSeek recently launched the DeepSeek R1 model, which has reached the first tier in terms of reasoning capabilities. However, DeepSeek R1 may still not be able to match Claude 3.5 Sonnet in the output of some daily tasks. A recent study by the Aider team showed that the best results can be achieved by adopting DeepSeek R1 + Claude 3.5 Sonnet. Inspired by this project, this project was completely rewritten through fastAPI. After 15 days of real testing by a large number of community users, we created some new combination usage plans. 1. Programming: Recommended to use deepclaude = deepseek r1 + claude 3.7 sonnet; 2. Content creation: Recommended to use deepgeminipro = deepseek r1 + gemini 2.0 pro (this solution can be used completely free); 3. Daily experiments: Recommended to use deepgeminiflash = deepseek r1 + gemini 2.0 flash (this solution can be used completely free). The project supports input and output in OpenAI-compatible formats, supports DeepSeek official APIs and third-party hosted APIs, and also supports Claude official APIs and transit APIs in the generation part, and provides special support for other Models in OpenAI-compatible formats. 🔥Recommended usage methods: 1. Users can run on their own servers, provide open API interfaces to the outside world, and access OneAPI to achieve unified distribution. 2. Connect to your daily big language model conversation and chat. For the first time, please open https://deepclaude. <device name>.heiyu.space/config for configuration, the default api key is 123456
