Using your Ollama models in Nightjar
Short answer
Nightjar scans Ollama's model directory at startup and lists what it finds. No export, no conversion, no second download — Ollama stores ordinary GGUF files.
How it works
At startup Nightjar looks in Ollama's standard storage location and reads the GGUF metadata of anything there, so sizing works exactly as it does for bundled models.
Ollama does not need to be running. The files are what matter.
If nothing appears
Confirm you have models with ollama list. Then check whether OLLAMA_MODELS points somewhere custom — if it does, add that folder in Nightjar's picker directly.
The scan happens at startup, so restart Nightjar after pulling something new.
Using both
They coexist happily. Ollama for scripting and local API work, Nightjar for reading the page and the documents in front of you, both off the same files on the same disk.
Other ways people ask this
- ollama browser
- use ollama models in browser
- I already have models downloaded in Ollama. Can a browser use them directly?