← Back to community benchmarks

gemma-4-26B-A4B-it-oQ8

M2 Max (38c) · 96 GB · 8bit · 2026-06-04
Performance
64k
tokens
652.0
PP tok/s
26.1
TG tok/s
100521
TTFT (ms)
29.2
Peak mem (GB)
Hardware
Chip M2 Max (38c)
Memory 96 GB
GPU Cores 38
Software
oMLX v0.4.1
macOS macOS 15.7.5
Context 65,536
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 907.4 51.6 25.8 GB view
4k 1,029 47.7 26.2 GB view
8k 1,034 46.2 26.4 GB view
16k 882.2 39.0 26.7 GB view
32k 812.5 33.5 27.6 GB view
64k 652.0 26.1 29.2 GB current
Batching Results
Batch Size TG tok/s Speedup
51.6 1.00×
65.0 1.26×
83.7 1.62×
161.4 3.13×