← Back to community benchmarks

Qwen3.5-35B-A3B4bit-MTP

M5 Max (40c) · 128 GB · 4bit · 2026-04-02
Performance
32k
tokens
3,082
PP tok/s
85.5
TG tok/s
10631
TTFT (ms)
22.5
Peak mem (GB)
Hardware
Chip M5 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.0
macOS macOS 26.4
Context 32,768
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,698 131.8 19.2 GB view
4k 3,424 119.6 20.2 GB view
32k 3,082 85.5 22.5 GB current
Batching Results
Batch Size TG tok/s Speedup
131.8 1.00×
201.6 1.53×
250.7 1.90×
360.9 2.74×