← Back to community benchmarks

Llama-3.2-1B-Instruct

M3 (10c) · 24 GB · 4bit · 2026-03-09
Performance
64k
tokens
278.2
PP tok/s
8.1
TG tok/s
235582
TTFT (ms)
3.3
Peak mem (GB)
Hardware
Chip M3 (10c)
Memory 24 GB
GPU Cores 10
Software
oMLX v0.2.6
macOS macOS 26.3
Context 65,536
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,249 102.4 1.3 GB view
4k 1,216 76.7 1.4 GB view
8k 1,130 59.3 1.5 GB view
16k 898.6 57.1 1.6 GB view
32k 596.2 32.2 2.2 GB view
64k 278.2 8.1 3.3 GB current
Batching Results
Batch Size TG tok/s Speedup
102.4 1.00×
67.9 0.66×
95.4 0.93×
140.2 1.37×