← Back to community benchmarks

Nanonets-OCR2-3B

M3 Max (40c) · 128 GB · unknown · 2026-03-09
Performance
32k
tokens
1,049
PP tok/s
35.2
TG tok/s
31237
TTFT (ms)
8.8
Peak mem (GB)
Hardware
Chip M3 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.2.6
macOS macOS 26.3
Context 32,768
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,905 51.0 7.5 GB view
4k 1,949 49.5 7.8 GB view
8k 1,802 46.8 7.8 GB view
16k 1,520 41.9 8.2 GB view
32k 1,049 35.2 8.8 GB current
Batching Results
Batch Size TG tok/s Speedup
51.0 1.00×
31.3 0.61×
47.3 0.93×
76.7 1.50×