← Back to community benchmarks

gemma-4-26B-A4B-it-uncensored-heretic-ara8bit-int8-affine

M3 Ultra (60c) · 256 GB · 8bit · 2026-06-02
Performance
32k
tokens
1,499
PP tok/s
57.2
TG tok/s
21863
TTFT (ms)
28.9
Peak mem (GB)
Hardware
Chip M3 Ultra (60c)
Memory 256 GB
GPU Cores 60
Software
oMLX v0.3.12
macOS macOS 26.4.1
Context 32,768
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,290 75.0 27.2 GB view
4k 1,650 74.3 27.6 GB view
8k 1,709 71.3 27.7 GB view
16k 1,647 65.6 28.1 GB view
32k 1,499 57.2 28.9 GB current
64k 1,206 46.7 30.6 GB view
128k 872.9 31.0 34.2 GB view
195k 646.1 23.6 37.9 GB view
Batching Results
Batch Size TG tok/s Speedup
75.0 1.00×
118.8 1.58×
162.1 2.16×
224.3 2.99×