← Back to community benchmarks

Nemotron-3-Nano-30B-A3B-RotorQuant

M5 Pro (20c) · 64 GB · 4bit · 2026-05-04
Performance
64k
tokens
4,063
PP tok/s
96.4
TG tok/s
16129
TTFT (ms)
21.0
Peak mem (GB)
Hardware
Chip M5 Pro (20c)
Memory 64 GB
GPU Cores 20
Software
oMLX v0.3.8
macOS macOS 26.4.1
Context 65,536
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,527 106.9 20.0 GB view
4k 1,950 104.0 21.1 GB view
8k 1,991 100.2 21.1 GB view
16k 4,445 105.0 20.9 GB view
32k 4,544 102.5 21.0 GB view
64k 4,063 96.4 21.0 GB current
128k 3,183 90.1 22.5 GB view
195k 2,607 83.0 25.1 GB view
Batching Results
Batch Size TG tok/s Speedup
106.9 1.00×
137.6 1.29×
181.0 1.69×
352.2 3.29×