← Back to community benchmarks

Qwen3.5-122B-A10B-5bit

M3 Max (40c) · 128 GB · 5bit · 2026-04-04
Performance
16k
tokens
1,238
PP tok/s
11.4
TG tok/s
13232
TTFT (ms)
84.1
Peak mem (GB)
Hardware
Chip M3 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 428.2 41.3 80.5 GB view
4k 483.7 39.4 82.0 GB view
8k 459.4 37.2 82.6 GB view
16k 1,238 11.4 84.1 GB current
32k 1,275 11.3 85.5 GB view
64k 1,143 10.6 87.4 GB view
128k 814.6 10.1 90.9 GB view
195k 688.0 9.8 94.9 GB view