Kimi K3 2800B (104B active)
on M5 Pro 64GB · WARP · 4,096 ctx
Summary
User reports Kimi K3 2.78T running at 0.45-0.62 tok/s on a 64 GB MacBook Pro with M5 Pro, streaming experts from internal NVMe.
Setup is the WARP engine with VQ3R 3-bit expert quantization, a 46.39 GB memory budget including a 17.56 GB expert cache, and a 982 GB converted container.
The 64-token run is a range; the 200-token column is absent because one run takes ten minutes and reads 4.6 TB. The engine also runs DeepSeek-V4.1-Flash 552B at 3.77/3.71 tok/s and GLM-5.3-Flash 313B at 3.32/3.86 tok/s on the same machine.