← Back to community benchmarks

supergemma4-26b-uncensored4bit-v2

M5 Max (40c) · 128 GB · 4bit · 2026-05-29
Performance
128k
tokens
1,389
PP tok/s
40.7
TG tok/s
94368
TTFT (ms)
22.4
Peak mem (GB)
Hardware
Chip M5 Max (40c)
Memory 128 GB
GPU Cores 40
Software
oMLX v0.3.12
macOS macOS 26.5
Context 131,072
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,576 121.5 14.0 GB view
4k 2,701 116.7 14.7 GB view
8k 3,546 110.8 14.8 GB view
16k 3,269 101.2 15.3 GB view
32k 2,742 83.0 16.2 GB view
64k 2,136 60.2 18.3 GB view
128k 1,389 40.7 22.4 GB current
Batching Results
Batch Size TG tok/s Speedup
121.5 1.00×
170.0 1.40×
234.0 1.93×
269.6 2.22×