← Back to community benchmarks

gemma-4-26b-a4b-it

M5 Pro (20c) · 64 GB · 4bit · 2026-07-21
Performance
16k
tokens
2,044
PP tok/s
62.3
TG tok/s
8014
TTFT (ms)
16.0
Peak mem (GB)
Hardware
Chip M5 Pro (20c)
Memory 64 GB
GPU Cores 20
Software
oMLX v0.5.1
macOS macOS 26.5.2
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 2,276 78.4 14.0 GB view
4k 2,218 74.9 14.5 GB view
8k 2,149 69.8 15.3 GB view
16k 2,044 62.3 16.0 GB current
32k 1,806 51.2 18.3 GB view
64k 1,339 37.5 18.3 GB view
128k 884.1 23.0 23.3 GB view
195k 619.2 17.2 27.4 GB view
Batching Results
Batch Size TG tok/s Speedup
78.4 1.00×
120.1 1.53×
160.9 2.05×
305.0 3.89×