← Back to community benchmarks

Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled-mxfp8

M1 Max (32c) · 64 GB · 8bit · 2026-03-10
Performance
32k
tokens
86.4
PP tok/s
8.0
TG tok/s
379197
TTFT (ms)
30.9
Peak mem (GB)
Hardware
Chip M1 Max (32c)
Memory 64 GB
GPU Cores 32
Software
oMLX v0.2.6
macOS macOS 26.3
Context 32,768
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 89.3 11.1 27.5 GB view
4k 91.8 10.7 27.9 GB view
8k 91.3 10.8 28.3 GB view
16k 89.5 9.0 29.2 GB view
32k 86.4 8.0 30.9 GB current
Batching Results
Batch Size TG tok/s Speedup
11.1 1.00×
15.2 1.37×
16.1 1.45×
18.3 1.65×