Performance
32k
tokens
194.3
PP tok/s
16.0
TG tok/s
168632
TTFT (ms)
8.9
Peak mem (GB)
Hardware
Chip
M4 (10c)
Memory
24 GB
GPU Cores
10
Software
oMLX
v0.3.4
macOS
macOS 26.2
Context
32,768