← Back to community benchmarks

Qwen3.5-2B

M1 Ultra (48c) · 64 GB · 4bit · 2026-03-11
Performance
4k
tokens
1,883
PP tok/s
168.6
TG tok/s
2176
TTFT (ms)
2.8
Peak mem (GB)
Hardware
Chip M1 Ultra (48c)
Memory 64 GB
GPU Cores 48
Software
oMLX v0.2.7
macOS macOS 26.3.1
Context 4,096
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,241 180.7 2.7 GB view
4k 1,883 168.6 2.8 GB current
8k 2,094 158.2 2.8 GB view
16k 2,019 132.6 3.0 GB view
32k 1,864 90.0 3.7 GB view
64k 1,570 55.8 5.1 GB view
Batching Results
Batch Size TG tok/s Speedup
180.7 1.00×
253.2 1.40×
322.9 1.79×
384.8 2.13×