← Back to community benchmarks

gemma-4-26b-a4b-it

M5 Max (40c) · 128 GB · 4bit · 2026-04-03
Performance
32k
tokens
3,236
PP tok/s
45.5
TG tok/s
10125
TTFT (ms)
17.9
Peak mem (GB)
Hardware
Chip M5 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.2
macOS macOS 26.4
Context 32,768
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 3,351 74.5 16.1 GB view
4k 3,686 71.4 16.5 GB view
8k 3,721 66.4 16.6 GB view
16k 3,635 58.5 17.1 GB view
32k 3,236 45.5 17.9 GB current
64k 2,539 30.7 19.6 GB view
128k 1,702 18.0 23.1 GB view
Batching Results
Batch Size TG tok/s Speedup
74.5 1.00×
99.6 1.34×
164.5 2.21×