← Back to community benchmarks

gemma-4-26b-a4b-it

M3 Ultra (80c) · 512 GB · bf16 · 2026-05-27
Performance
16k
tokens
2,016
PP tok/s
54.9
TG tok/s
8128
TTFT (ms)
49.1
Peak mem (GB)
Hardware
Chip M3 Ultra (80c)
Memory 512 GB
GPU Cores 80
Software
oMLX v0.3.10
macOS macOS 26.4.1
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,259 62.9 47.8 GB view
4k 1,750 61.3 48.5 GB view
8k 2,061 59.4 48.8 GB view
16k 2,016 54.9 49.1 GB current
32k 1,817 48.9 50.0 GB view
64k 1,485 38.5 52.0 GB view
128k 1,056 26.6 56.2 GB view
195k 789.4 20.2 61.2 GB view
Batching Results
Batch Size TG tok/s Speedup
62.9 1.00×
63.0 1.00×
106.0 1.69×
79.5 1.26×