← Back to community benchmarks

gemma-4-E4B-it

M3 Max (40c) · 128 GB · 8bit · 2026-05-03
Performance
195k
tokens
543.2
PP tok/s
6.3
TG tok/s
368172
TTFT (ms)
34.9
Peak mem (GB)
Hardware
Chip M3 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.7
macOS macOS 26.4.1
Context 200,000
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,641 41.6 9.0 GB view
4k 1,851 42.6 9.1 GB view
8k 1,997 41.2 9.1 GB view
16k 1,767 37.1 9.3 GB view
32k 1,441 32.1 9.9 GB view
64k 1,052 18.8 10.5 GB view
128k 492.8 7.6 33.6 GB view
195k 543.2 6.3 34.9 GB current
Batching Results
Batch Size TG tok/s Speedup
41.6 1.00×
60.3 1.45×