In versions 2.0 and 2.1.0, are only OpenAI and Gemini supported, and other models not supported? For example, Deepseek and Qwen?

[mozou]

I searched the issues and found previous records about deepseek and qwen. However, when I searched for keywords related to "qwen" and "deepseek" in the code, I found nothing.

The code is still configured with OpenAI and Gemini.

[NA-Wen]

In the current configuration, our model provider refer to the API interface, not specific model type.

Both DeepSeek and Qwen expose OpenAI-compatible APIs. So if you are using DeepSeek or Qwen via an OpenAI-compatible endpoint, you can simply choose OpenAI as model provider and configure the corresponding base URL and API key.

[mozou]

@NA-Wen Thanks for your reply. In version 2.1.0, I was able to get it to run smoothly after modifying the model in the workflow. I discovered that Deepseek doesn't seem to support /response, which caused a 404 error.

2026-02-12 16:30:51,561 - httpx - INFO - HTTP Request: POST https://api.deepseek.com/responses "HTTP/1.1 404 Not Found"

[zxrys]

Hello. Please set BASE_URL to https://api.deepseek.com:

It works: