Ollama integration issue.

[ytchuan]

I am not a programmer but interested to see how it works. I have since trying to config it to run on local ollama. here's what I do.

1. Running windows 11 pro, local Docker for dekstop, Ollama for windows (not in Docker), installed AgentZero as per instructed.
2. Setting.Chat model+Utility Model+Webbrowser : changed to Ollama and Gemma3:12b (or others eg deepseek)
3. Embedding model to Ollama+mxbai-embed-large:latest
4. CMD prompt on windows, Ollama run Gemma3:12b (generally won't work if I don't do this step, any auto start will be grateful).
5. at times set up like this manage to run for a while, but invariably run into error with red error screens.

My questions also inc.

• if embed model is also from Ollama, I can't 'run' the model before hand since its meant for embedding. There seems to have issue with running emdedded model from Ollama.
• Error code last line appear to show it tried to run ollama via different port instead of the fixed 11434.
• I have also add additional script on .env file to fix the ollama port to 11434 according to another message poster earlier. It does not work. My Ollama is installed on windows direct, not through Docker. ( from Ollama integrations #357 , where I added OLLAMA_BASE_URL="http://host.docker.internal:11434" within the .env file) Does not work for me.

Any idea will be great.