← Back to community benchmarks

gemma-4-26b

M3 Pro (18c) · 36 GB · 4bit · 2026-05-06
Performance
64k
tokens
338.7
PP tok/s
21.2
TG tok/s
193469
TTFT (ms)
17.7
Peak mem (GB)
Hardware
Chip M3 Pro (18c)
Memory 36 GB
GPU Cores 18
Software
oMLX v0.3.8
macOS macOS 26.4.1
Context 65,536
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 433.8 42.2 14.3 GB view
4k 503.6 40.1 14.9 GB view
8k 517.4 38.6 15.1 GB view
16k 484.5 29.8 15.2 GB view
32k 416.0 27.9 16.0 GB view
64k 338.7 21.2 17.7 GB current
Batching Results
Batch Size TG tok/s Speedup
42.2 1.00×
61.1 1.45×
69.7 1.65×
84.1 1.99×