← Back to community benchmarks

Qwen3.5-122B-A10B-oQ4

M5 Max (40c) · 128 GB · 4bit · 2026-03-26
Performance
8k
tokens
1,052
PP tok/s
16.0
TG tok/s
7787
TTFT (ms)
74.1
Peak mem (GB)
Hardware
Chip M5 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.2.20
macOS macOS 26.4
Context 8,192
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 821.1 16.5 69.9 GB view
4k 1,072 16.3 72.9 GB view
8k 1,052 16.0 74.1 GB current
Batching Results
Batch Size TG tok/s Speedup
16.5 1.00×
29.8 1.81×
51.6 3.13×
76.5 4.64×