← Back to community benchmarks

gemma-3-27b-it-qat

M3 Ultra (80c) · 512 GB · 4bit · 2026-03-10
Performance
1k
tokens
397.3
PP tok/s
20.3
TG tok/s
2580
TTFT (ms)
16.8
Peak mem (GB)
Hardware
Chip M3 Ultra (80c)
Memory 512 GB
GPU Cores 80
Software
oMLX v0.2.7
macOS macOS 15.7.4
Context 1,024
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 397.3 20.3 16.8 GB current
4k 401.6 31.1 17.7 GB view
8k 401.3 27.6 18.0 GB view
16k 400.0 23.2 18.6 GB view
32k 390.2 18.7 19.9 GB view
64k 370.0 12.7 22.5 GB view
Batching Results
Batch Size TG tok/s Speedup
20.3 1.00×
56.7 2.79×
74.4 3.67×
85.8 4.23×