← Back to community benchmarks

gemma-4-26b-a4b-it

M5 Max (40c) · 128 GB · 6bit · 2026-04-06
Performance
16k
tokens
3,254
PP tok/s
69.2
TG tok/s
5035
TTFT (ms)
20.9
Peak mem (GB)
Hardware
Chip M5 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.4
macOS macOS 26.4
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 2,001 93.6 20.0 GB view
4k 3,125 85.3 20.4 GB view
8k 3,415 79.0 20.5 GB view
16k 3,254 69.2 20.9 GB current
32k 3,022 54.9 21.7 GB view
64k 2,321 36.6 23.4 GB view
128k 1,614 18.6 27.0 GB view
195k 1,224 11.2 30.6 GB view
Batching Results
Batch Size TG tok/s Speedup
93.6 1.00×
134.3 1.43×
185.7 1.98×
228.7 2.44×