← Back to community benchmarks

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

M5 Pro (20c) · 64 GB · 6bit · 2026-04-05
Performance
16k
tokens
414.2
PP tok/s
3.4
TG tok/s
39554
TTFT (ms)
30.0
Peak mem (GB)
Hardware
Chip M5 Pro (20c)
Memory 64 GB
GPU Cores 20
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 390.6 12.6 22.1 GB view
4k 425.3 12.5 23.6 GB view
8k 426.4 7.4 24.3 GB view
16k 414.2 3.4 30.0 GB current
32k 385.6 10.0 34.2 GB view
64k 325.1 9.5 33.1 GB view
Batching Results
Batch Size TG tok/s Speedup
12.6 1.00×
22.2 1.76×