Hi n8n team,
I’m Jay (Nguyen Thieu Toan), an Verified n8n Creator, and I’d like to raise a concern that is starting to affect quite a few users and clients who are trying to adopt DeepSeek inside n8n for production AI Agent workflows.
Right now, whenever DeepSeek Chat Model is used with AI Agent + tools, we consistently run into the error “The reasoning_content in the thinking mode must be passed back to the API.” From what I can see in recent threads and GitHub issues, this has already been acknowledged as a known limitation in how reasoning_content is handled when the agent calls tools, but there hasn’t been a visible fix shipped yet. In practice, this means that for serious agent use cases (where tools are essential), DeepSeek is effectively not a viable option inside the official n8n nodes.
What worries me is that the community is starting to rely on unofficial DeepSeek nodes or custom patches just to work around this, which solves the error for some people but introduces new risks: inconsistent behavior, harder maintenance, and less confidence for teams that want to standardize on n8n as their main automation and AI platform. This isn’t just a “nice to have” anymore; DeepSeek has become a real alternative that many builders and clients actively prefer over OpenAI or Gemini/Vertex, and the current limitation is blocking them from using it properly with agents and tools.
Could you please share an update on the status of the official fix for DeepSeek + AI Agent tool calls (the reasoning_content handling), and whether there is an estimated release window we can communicate to our users and clients? Even a rough indication like “planned and in progress for an upcoming minor release” would already help a lot of us decide whether to wait for the official implementation or keep investing in community‑level workarounds.
Thanks in advance for any clarity you can provide here on behalf of the growing group of n8n users who want to use DeepSeek in a stable, first‑class way.
