Kimi K3.1 locally on one GPU 2 points by Argonautlabs 1 day ago 2 tokens/s decode and 40 tokens/s prefillhttps://argonautlabs.ai/
Argonautlabs 1 day ago With a drafter decode speed is 2.64 tok/sec.Mac version should be released next week.
With a drafter decode speed is 2.64 tok/sec.
Mac version should be released next week.