← Back to community benchmarks

gemma-4-26b-a4b-it

M3 Ultra (80c) · 512 GB · 4bit · 2026-04-16
Performance
195k
tokens
813.2
PP tok/s
7.8
TG tok/s
245934
TTFT (ms)
24.9
Peak mem (GB)
Hardware
Chip M3 Ultra (80c)
Memory 512 GB
GPU Cores 80
Software
oMLX v0.3.5.dev1
macOS macOS 26.1
Context 200,000
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,238 98.8 14.2 GB view
4k 1,857 93.6 14.7 GB view
8k 2,224 81.0 14.8 GB view
16k 2,172 66.9 15.2 GB view
32k 1,967 49.7 16.0 GB view
64k 1,585 26.1 17.7 GB view
128k 1,099 12.2 21.2 GB view
195k 813.2 7.8 24.9 GB current
Batching Results
Batch Size TG tok/s Speedup
98.8 1.00×
158.7 1.61×
231.3 2.34×