← Back to community benchmarks

Qwen3-32B

M3 Ultra (60c) · 96 GB · 4bit · 2026-03-15
Performance
4k
tokens
241.6
PP tok/s
28.1
TG tok/s
16951
TTFT (ms)
18.0
Peak mem (GB)
Hardware
Chip M3 Ultra (60c)
Memory 96 GB
GPU Cores 60
Software
oMLX v0.2.7
macOS macOS 26.2
Context 4,096
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 246.4 32.5 17.0 GB view
4k 241.6 28.1 18.0 GB current
Batching Results
Batch Size TG tok/s Speedup
32.5 1.00×
49.4 1.52×
56.4 1.74×