← Back to community benchmarks

Qwen3.5-122B-A10B

M5 Max (40c) · 128 GB · 4bit · 2026-04-01
Performance
16k
tokens
1,430
PP tok/s
54.3
TG tok/s
11461
TTFT (ms)
68.9
Peak mem (GB)
Hardware
Chip M5 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.0
macOS macOS 26.3.2
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 922.9 65.8 65.6 GB view
4k 1,456 62.7 67.0 GB view
8k 1,542 60.2 67.7 GB view
16k 1,430 54.3 68.9 GB current
32k 1,226 42.7 71.2 GB view
64k 886.6 31.4 76.1 GB view
128k 555.6 20.2 85.7 GB view
195k 394.2 14.4 95.9 GB view
Batching Results
Batch Size TG tok/s Speedup
65.8 1.00×
99.9 1.52×
124.7 1.90×