Well goose can use a variety of backends but i havent had much success with small local models
Discussion
well that's very gay, because i can run a qwen 14B model on my 16gb 256bit RX 7800XT that spits out text as fast as a cloud machine does on my subscription. i know this model can work with tools so i know it can work with an agent controller.
i'm just gonna hold my beer until jetbrains opens up configuring arbitrary backends on junie because i'm pretty damn sure that it will be competitive with claude 3.7