← Back to community benchmarks

gemma-4-26B-A4B-it

M5 Pro (20c) · 64 GB · 6bit · 2026-05-24
Performance
16k
tokens
1,763
PP tok/s
49.7
TG tok/s
9292
TTFT (ms)
21.3
Peak mem (GB)
Hardware
Chip M5 Pro (20c)
Memory 64 GB
GPU Cores 20
Software
oMLX v0.3.9
macOS macOS 26.5
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,417 59.3 20.0 GB view
4k 1,582 57.2 20.7 GB view
8k 1,799 55.1 20.8 GB view
16k 1,763 49.7 21.3 GB current
32k 1,587 42.6 22.2 GB view
64k 1,237 31.9 24.3 GB view
128k 767.3 20.1 28.4 GB view
195k 584.6 15.2 33.4 GB view
Batching Results
Batch Size TG tok/s Speedup
59.3 1.00×
26.5 0.45×
109.1 1.84×
129.1 2.18×