1
When Loading the model make sure you check 'Manually Choose Model Load Parameter' for developer options, and also 'advanced settings'
2. Context Size / GPU / LOAD MODEL
Lower 'Context' if only using model for chat, 'Higher' Using tools & Agents!
Increase GPU Offload as much as you can, Use models that fit based on ram, (My new mac is 32GB) (old macs 16GB)
click load
when model loads start a chat, sometime you have to click the model again