← Back to community benchmarks

Qwen3.5-27B-heretic

M3 Ultra (60c) · 96 GB · 8bit · 2026-03-12
Performance
64k
tokens
253.9
PP tok/s
13.8
TG tok/s
258078
TTFT (ms)
36.2
Peak mem (GB)
Hardware
Chip M3 Ultra (60c)
Memory 96 GB
GPU Cores 60
Software
oMLX v0.2.8
macOS macOS 26.3.1
Context 65,536
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 313.9 22.7 29.1 GB view
4k 318.7 22.1 29.5 GB view
16k 305.7 20.1 30.8 GB view
32k 287.5 18.2 32.6 GB view
64k 253.9 13.8 36.2 GB current
Batching Results
Batch Size TG tok/s Speedup
22.7 1.00×
39.0 1.72×
49.5 2.18×
57.3 2.52×