← Back to community benchmarks

Qwen3.5-2B

M3 Max (40c) · 48 GB · 6bit · 2026-04-18
Performance
8k
tokens
3,121
PP tok/s
155.6
TG tok/s
2625
TTFT (ms)
3.1
Peak mem (GB)
Hardware
Chip M3 Max (40c)
Memory 48 GB
GPU Cores 40
Software
oMLX v0.3.6
macOS macOS 26.3.1
Context 8,192
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 2,001 173.0 2.5 GB view
4k 2,886 166.3 2.8 GB view
8k 3,121 155.6 3.1 GB current
16k 3,078 138.6 3.2 GB view
Batching Results
Batch Size TG tok/s Speedup
173.0 1.00×
291.5 1.68×
344.4 1.99×
384.1 2.22×