← Back to community benchmarks

Carnice-V2-27b

M5 Max (40c) · 128 GB · 4bit · 2026-05-20

Performance

32k

tokens

595.2

PP tok/s

26.3

TG tok/s

55050

TTFT (ms)

22.8

Peak mem (GB)

Hardware

Chip M5 Max (40c)

Memory 128 GB

GPU Cores 40

Software

oMLX v0.3.8

macOS macOS 26.4.1

Context 32,768

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	799.1	128.3	17.2 GB	view
4k	634.5	32.7	17.3 GB	view
8k	757.6	31.5	18.3 GB	view
16k	660.6	29.4	19.8 GB	view
32k	595.2	26.3	22.8 GB	current
64k	408.0	11.2	28.8 GB	view
128k	217.8	9.7	41.2 GB	view
195k	158.8	7.9	55.9 GB	view