← Back to community benchmarks

gemma-3-4b-it

M1 Ultra (48c) · 128 GB · 4bit · 2026-03-10
Performance
16k
tokens
909.1
PP tok/s
64.7
TG tok/s
18024
TTFT (ms)
4.6
Peak mem (GB)
Hardware
Chip M1 Ultra (48c)
Memory 128 GB
GPU Cores 48
Software
oMLX v0.2.6
macOS macOS 26.3.1
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 926.5 94.7 4.0 GB view
4k 886.0 90.6 4.2 GB view
8k 922.8 78.4 4.3 GB view
16k 909.1 64.7 4.6 GB current
32k 878.1 50.2 5.2 GB view
64k 805.0 32.6 6.4 GB view
Batching Results
Batch Size TG tok/s Speedup
94.7 1.00×
131.4 1.39×
163.6 1.73×