← Back to community benchmarks

Llama-3.2-3B-Instruct

M4 (10c) · 24 GB · 4bit · 2026-03-09
Performance
4k
tokens
478.6
PP tok/s
37.0
TG tok/s
8561
TTFT (ms)
2.6
Peak mem (GB)
Hardware
Chip M4 (10c)
Memory 24 GB
GPU Cores 10
Software
oMLX v0.2.6
macOS macOS 15.7.4
Context 4,096
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 520.4 45.2 2.3 GB view
4k 478.6 37.0 2.6 GB current
Batching Results
Batch Size TG tok/s Speedup
45.2 1.00×
83.7 1.85×
99.4 2.20×