← Back to community benchmarks

Qwen3.5-27B

M3 Ultra (80c) · 512 GB · 8bit · 2026-03-09
Performance
4k
tokens
421.8
PP tok/s
21.0
TG tok/s
9711
TTFT (ms)
29.5
Peak mem (GB)
Hardware
Chip M3 Ultra (80c)
Memory 512 GB
GPU Cores 80
Software
oMLX v0.2.6
macOS macOS 26.3.1
Context 4,096
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 413.0 21.6 29.1 GB view
4k 421.8 21.0 29.5 GB current
8k 417.0 20.4 29.9 GB view
32k 378.9 17.3 32.6 GB view
64k 333.7 15.0 36.2 GB view
Batching Results
Batch Size TG tok/s Speedup
21.6 1.00×
38.2 1.77×
56.9 2.63×
70.0 3.24×