← Back to community benchmarks

gemma-4-26b-a4b-it

M5 Max (40c) · 128 GB · 8bit · 2026-04-03
Performance
64k
tokens
1,966
PP tok/s
15.8
TG tok/s
33341
TTFT (ms)
30.3
Peak mem (GB)
Hardware
Chip M5 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.2
macOS macOS 26.4
Context 65,536
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 2,326 64.4 26.8 GB view
4k 3,201 59.6 27.2 GB view
8k 3,197 55.2 27.4 GB view
16k 2,829 47.8 27.8 GB view
32k 2,416 29.7 28.6 GB view
64k 1,966 15.8 30.3 GB current
Batching Results
Batch Size TG tok/s Speedup
64.4 1.00×
85.9 1.33×
125.9 1.95×
166.1 2.58×