← Back to community benchmarks

LFM2-8B-A1B-3bit

M4 (10c) · 16 GB · 3bit · 2026-03-12
Performance
1k
tokens
813.5
PP tok/s
101.8
TG tok/s
1260
TTFT (ms)
5.1
Peak mem (GB)
Hardware
Chip M4 (10c)
Memory 16 GB
GPU Cores 10
Software
oMLX v0.2.9
macOS macOS 26.3.1
Context 1,024
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 813.5 101.8 5.1 GB current
4k 817.7 93.7 5.2 GB view
Batching Results
Batch Size TG tok/s Speedup
101.8 1.00×
131.3 1.29×
141.0 1.39×