Using fastmcp with ollama: qwen, mistral, llama, ....

[jordimassaguerpla]

Hi!

We are implementing an MCP Server with FastMCP. We use openWebUI with the mcpo proxy. All works well with GPT or Gemini. However, when we try to connect it to Qwen or Mistral, which ran with ollama, it does not work, meaning it does not call the tools to get the answers... I see there is docs for Gemini, Claude and GPT, but not for Qwen or Llama or Mistral... https://github.com/jlowin/fastmcp/tree/main/docs/integrations

Is running models with ollama not supported with FastMCP?

[jbrodo]

Have you tried this https://github.com/jonigl/mcp-client-for-ollama ?