Local model problems: Ollama not found, an empty list, a machine that is not reachable

The Ollama row in Settings, Intelligence, API keys tells you which of five states you are in and what to do next, whether Ollama is on this PC or another computer.

Last checked against the app on · View as Markdown

Steps

  1. Open Settings, then Intelligence, then API keys, and read the line on the Ollama row. It names the address it is using after "Reached at".
  2. "Ollama is not running": nothing answers at that address. Start Ollama or install it from ollama.com. If Ollama is on another computer, the address is wrong, or that machine is not listening on the network. See the guide on Ollama on another machine.
  3. "Ollama has no models installed": Ollama answers but has nothing pulled. In a terminal on the machine that runs Ollama, run "ollama pull" with a model name, as the app’s own hint shows.
  4. "No local model is chosen": Ollama answers at that address but no model is set. Choose one in the model button in Chat.
  5. A line saying a named model is not installed: pull that model with "ollama pull", or choose one you have.
  6. "Ollama did not answer properly": something answers at the address but gave no model list. Check the address points at Ollama and not at another program.

How to tell it worked

The Ollama row says "Ollama is answering" and gives the model and address.

If it does not work

The row says "Ollama has no address".
No base address is set on this PC. Put one in the .neuralvault-ollama.json file in your user folder, as the guide on Ollama on another machine shows.
Models are installed but the picker is empty.
Check the row first. If it says Ollama is answering, see the guide on a model missing from the picker.
It works on the other machine itself but not from your PC.
Then Ollama is only listening on that machine. See the guide on Ollama on another machine, and the security note before you widen it.