My current setup > pi > unsloth/Qwen3.6-35B-A3B-GGUF > llamacpp what i got > ~50 tok/s > get most of the work done without any issue, can run autonomously for hours I am happy with this setup, but i wonder if there is a better setup/ model?
