Ollama recognized but no LLMs listed

[devChrisStudios]

I have had this problem using Ollama with other things as well but when I try to connect port 11434 and it works but none of my models show up.

[darekwojciechowski]

"Port connects but the list is empty" almost always means one of two things, and one command tells you which.

Run this from the same machine the app is on:

curl http://localhost:11434/api/tags

If you get back {"models":[]}, the Ollama that's answering genuinely has no models. That happens when models were pulled under a different user than the one running the service — ollama pull as your user writes to ~/.ollama, but the systemd service reads /usr/share/ollama/.ollama. Pull once more as the service user and they'll show up.

If you get a real list of models but the app still shows nothing, it's CORS. Ollama only answers browser requests from origins it's been told to trust, and a version check can slip through where the model list doesn't. Set:

OLLAMA_ORIGINS=*
OLLAMA_HOST=0.0.0.0:11434

On Linux with systemd:

systemctl edit ollama

and add:

[Service]
Environment="OLLAMA_HOST=0.0.0.0:11434"
Environment="OLLAMA_ORIGINS=*"

then systemctl daemon-reload && systemctl restart ollama.

On macOS: launchctl setenv OLLAMA_ORIGINS "*" and restart the Ollama app.

One thing worth ruling out first if the app runs in Docker: localhostinside a container is the container itself, not your host. There you'd want http://host.docker.internal:11434 instead.