← Back to community benchmarks

gemma-4-26b-a4b-it

M5 Max (40c) · 128 GB · 4bit · 2026-06-01
Performance
32k
tokens
3,240
PP tok/s
81.0
TG tok/s
10115
TTFT (ms)
22.9
Peak mem (GB)
Hardware
Chip M5 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.12
macOS macOS 26.4
Context 32,768
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 2,110 109.0 21.2 GB view
4k 3,454 106.8 21.6 GB view
8k 3,719 95.5 21.8 GB view
16k 3,622 93.8 22.2 GB view
32k 3,240 81.0 22.9 GB current
64k 2,436 61.6 24.6 GB view
128k 1,359 34.2 28.2 GB view
Batching Results
Batch Size TG tok/s Speedup
109.0 1.00×
217.2 1.99×
383.6 3.52×
786.7 7.22×