← Back to community benchmarks

Llama-3.2-1B-Instruct

M4 (10c) · 24 GB · 4bit · 2026-03-10
Performance
32k
tokens
476.1
PP tok/s
19.8
TG tok/s
68822
TTFT (ms)
2.2
Peak mem (GB)
Hardware
Chip M4 (10c)
Memory 24 GB
GPU Cores 10
Software
oMLX v0.2.6
macOS macOS 26.2
Context 32,768
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,135 97.6 1.3 GB view
4k 1,090 70.5 1.4 GB view
8k 1,013 55.5 1.5 GB view
16k 811.1 53.3 1.6 GB view
32k 476.1 19.8 2.2 GB current
64k 193.9 9.5 3.3 GB view
Batching Results
Batch Size TG tok/s Speedup
97.6 1.00×
86.1 0.88×
102.9 1.05×
110.5 1.13×