Kimi K3.1 locally on one GPU
2 tokens/s decode and 40 tokens/s prefill
https://argonautlabs.ai/
https://argonautlabs.ai/
Mac version is released.
2 tok/s decode 8.4 tok/s prefill
Its a source build with weights downloaded separately.
Code: https://github.com/argonautlabsai/argodrive-ds4-kimi-k3.1 Full results and setup: https://argonautlabs.ai/research/kimi-k31-single-gpu.html#ap...
Mac version should be released next week.