LM Studio works perfectly fine, because agent zero is running in Docker and LM studio people usually run on their computers When you launch LM Studio in develoer mode and you see that model is the model say reachable at this end point, and then you can copy that URL to put into agent 0. You don't need to use an API key for LM studio but in agent 0 you have to select LM studio in your setup. I'm a Windows guy and my Mac use is kind of limited but there's three keys which is the api endpoint for LM studio has to be in Agent zeros model information, it should work just right through the agent 0 interface, but if you have problems, when docker is running and you can see the files of agent 0, you have a .env file in the main directory and a conf/model_providers.yaml which is what is picked up by models.py so if your settings arent quite working you can manually edit the yaml file. Have your LLM in LM studio help you with the setup, is what I would do.