Add DeepSeek models

[kriskwiatkowski]

DeepSeek is way cheaper than other models. Would be interesting to add possibility to integrate DeepSeek.

[tmseidel]

@kriskwiatkowski DeepSeek has an OpenAI and Anthropic compatible Rest-API, so it already should work with the current feature-set, simply create an AI-Integration, choose as provider openAI and set the properties you can find on DeepSeek's API website -> https://api-docs.deepseek.com/

Would like to hear if that's working for you...

[kriskwiatkowski]

@tmseidel let me try
BTW: This project is really good stuff! Amazing help

[tmseidel]

@kriskwiatkowski I've tested the DeepSeek Models via the OpenAI Integration. The simple LLM requests and also the agentic requests with native tool calling were working like expected, so I'll close this issue.

[kriskwiatkowski]

@tmseidel sorry for late response. Yup, it all works all right ;)
I've tried with open-ai model.

[richardchen874-sys]

Good to know that DeepSeek works through the OpenAI integration, including agentic requests with native tool calling.

That’s useful confirmation, because for many teams the basic chat path is usually easy, but tool calling, streaming, retries, and production stability are where compatibility issues tend to appear.

For people using DeepSeek in real workflows, it would also be interesting to compare latency, token cost, and reliability against other OpenAI-compatible providers.

I’m looking into how indie devs and small AI teams use lower-cost models like DeepSeek / BytePlus in agent workflows, so this confirmation is helpful.