← Back to community benchmarks

Qwen3.5-122B-A10B-5bit

M5 Max (40c) · 128 GB · 5bit · 2026-04-05
Performance
16k
tokens
1,324
PP tok/s
47.5
TG tok/s
12378
TTFT (ms)
83.1
Peak mem (GB)
Hardware
Chip M5 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.2
macOS macOS 26.4
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,005 58.4 79.8 GB view
4k 1,440 55.8 81.3 GB view
8k 1,443 53.9 81.9 GB view
16k 1,324 47.5 83.1 GB current
32k 1,087 38.7 85.5 GB view
64k 831.8 28.1 90.3 GB view
128k 507.2 18.2 100.0 GB view
Batching Results
Batch Size TG tok/s Speedup
58.4 1.00×
77.0 1.32×
86.8 1.49×
105.4 1.80×