← Back to community benchmarks

gemma-4-26b-a4b-it

M4 Pro (20c) · 64 GB · 8bit · 2026-04-06
Performance
16k
tokens
692.4
PP tok/s
29.6
TG tok/s
23664
TTFT (ms)
26.7
Peak mem (GB)
Hardware
Chip M4 Pro (20c)
Memory 64 GB
GPU Cores 20
Software
oMLX v0.3.4
macOS macOS 26.4
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 643.5 45.1 25.7 GB view
4k 705.7 43.2 26.2 GB view
8k 726.5 39.9 26.3 GB view
16k 692.4 29.6 26.7 GB current
32k 605.0 20.9 27.5 GB view
64k 503.8 14.5 29.2 GB view
128k 370.5 8.1 32.7 GB view
195k 280.3 6.4 36.5 GB view
Batching Results
Batch Size TG tok/s Speedup
45.1 1.00×
65.2 1.45×
82.6 1.83×
105.8 2.35×