Can you provide steps to reproduce so I can see one of these deterministic LLMs running myself? API based or local models.

I’d recommend starting with llamafile due to its simplicity and then move up to llamacpp and Unsloth.