-------
Others may know this already, but it's a win for me because I did not.
I managed to get my local Ollama instance connected to Copilot Chat inside of VS Code.
Its a little slower than paid models but seems to behaves better than opencode connected the same Ollama instance. I don't understand why that would be but this is pretty interesting.
Amazing times we live in.
UPDATE: How I did it.
Prereq: running Ollama instance on local host
- Open Chat in vscode
- Click on Auto to Pick a model.
- On Language Model dialog, click + Add Models
- Select Ollama and fill in the info. My Ollama is on the same machine with default host & port
- Select the model you have available.
You can pin a model that you use a lot. Which is handy for a digital hoarder like me.