Feature request

ollama currently doesn't work very well when users just follow the officially documented way: https://github.com/yetone/avante.nvim/wiki/Custom-providers#ollama

It is currently treated as an OpenAI-compatible provider but as various bug reports show, this isn't really the case and ollama won't work as expected with avante: #998 #1067 #782 #910

The truth of the matter seems to be that the OpenAI-compatible endpoint isn't too useful for avante's use case, primarily because it doesn't support setting the context size.

To make things work somehow smoothly, you currently have to do this which is far from a turn-key solution.

So my feature suggestion is: Add a dedicated ollama provider to make ollama just work without going through hoops and adding 50 lines of config. It should use the ollama chat API.

Motivation

Using ollama with avante should be smooth, easy, and obvious. It currently isn't.

Other

No response