← Back to community benchmarks

Qwen3.5-9B-DeepSeek-V4-Flash-oQ4-mtp

M4 (10c) · 24 GB · 4bit · 2026-05-20

Performance

16k

tokens

145.7

PP tok/s

20.5

TG tok/s

112471

TTFT (ms)

8.0

Peak mem (GB)

Hardware

Chip M4 (10c)

Memory 24 GB

GPU Cores 10

Software

oMLX v0.3.9.dev2

macOS macOS 26.4.1

Context 16,384

Performance by Context Length

Context	PP tok/s	TG tok/s	Peak Mem
1k	176.3	27.3	6.0 GB	view
4k	176.4	26.8	6.6 GB	view
8k	170.0	24.9	7.1 GB	view
16k	145.7	20.5	8.0 GB	current
32k	124.3	17.4	9.8 GB	view