← Back to community benchmarks

Qwen3.5-27B-HereticoQ6

M1 Ultra (64c) · 128 GB · 6bit · 2026-03-24
Performance
16k
tokens
167.5
PP tok/s
2.7
TG tok/s
97834
TTFT (ms)
26.7
Peak mem (GB)
Hardware
Chip M1 Ultra (64c)
Memory 128 GB
GPU Cores 64
Software
oMLX v0.2.20
macOS macOS 26.4
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 169.9 21.8 23.2 GB view
4k 172.0 17.9 24.8 GB view
8k 168.9 14.0 25.4 GB view
16k 167.5 2.7 26.7 GB current
32k 158.9 3.0 29.2 GB view
Batching Results
Batch Size TG tok/s Speedup
21.8 1.00×
26.2 1.20×
27.6 1.27×