← Back to community benchmarks

gemma-4-26b-a4b-it

M5 Max (40c) · 128 GB · 4bit · 2026-05-18
Performance
128k
tokens
1,203
PP tok/s
33.2
TG tok/s
108964
TTFT (ms)
21.2
Peak mem (GB)
Hardware
Chip M5 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.8
macOS macOS 26.4.1
Context 131,072
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,790 113.2 14.3 GB view
4k 1,585 86.0 14.9 GB view
8k 1,850 89.6 15.1 GB view
16k 2,057 84.4 15.6 GB view
32k 2,088 70.2 16.0 GB view
64k 1,735 50.9 17.7 GB view
128k 1,203 33.2 21.2 GB current
195k 850.1 22.8 24.9 GB view
Batching Results
Batch Size TG tok/s Speedup
113.2 1.00×
124.6 1.10×
163.3 1.44×
201.4 1.78×