MLX and Core ML ports other people published. Latency numbers on those cards are theirs. This path only points at the right repo.
Walk this in order
- Decide whether you want Metal via MLX or a Core ML bundle aimed at the Neural Engine.
- Install the matching package from PyPI. Do not mix the two runtimes in one first experiment.
- Start from that author’s README and their reported short-decision times. This site did not re-run them.
- If you want a visible loop, open the Terminal Snake card and leave this site for the repo.
- When you need energy notes, the ANE write-up is a separate card from the runtime.
Leave this site when you need the source
Want a person on the call
The path above stays free. A walkthrough request is reviewed like a sponsor slot.