← Back to community benchmarks

gemma-4-26B-A4B-it

M3 Ultra (60c) · 96 GB · 8bit · 2026-05-29
Performance
195k
tokens
623.6
PP tok/s
22.2
TG tok/s
320714
TTFT (ms)
39.2
Peak mem (GB)
Hardware
Chip M3 Ultra (60c)
Memory 96 GB
GPU Cores 60
Software
oMLX v0.3.12
macOS macOS 26.5
Context 200,000
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,319 82.8 25.8 GB view
4k 1,652 79.9 26.4 GB view
8k 1,692 76.2 26.6 GB view
16k 1,625 70.4 27.1 GB view
32k 1,464 60.0 28.0 GB view
64k 1,137 45.3 30.0 GB view
128k 848.8 30.2 34.2 GB view
195k 623.6 22.2 39.2 GB current
Batching Results
Batch Size TG tok/s Speedup
82.8 1.00×
122.1 1.47×
168.5 2.04×
230.3 2.78×