← Back to community benchmarks

Qwen3.5-0.8B

M5 (10c) · 32 GB · 4bit · 2026-04-18
Performance
16k
tokens
3,064
PP tok/s
151.4
TG tok/s
5347
TTFT (ms)
2.5
Peak mem (GB)
Hardware
Chip M5 (10c)
Memory 32 GB
GPU Cores 10
Software
oMLX v0.3.6
macOS macOS 26.4.1
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 2,556 208.0 1.4 GB view
4k 2,662 187.1 1.6 GB view
8k 2,938 171.8 1.9 GB view
16k 3,064 151.4 2.5 GB current
32k 2,507 110.8 3.8 GB view
64k 1,649 59.3 4.3 GB view
128k 722.8 26.6 7.3 GB view
195k 482.0 18.1 11.9 GB view