← Back to community benchmarks

gemma-4-26b-a4b

M3 Ultra (60c) · 256 GB · bf16 · 2026-05-11
Performance
32k
tokens
1,416
PP tok/s
40.1
TG tok/s
23147
TTFT (ms)
50.5
Peak mem (GB)
Hardware
Chip M3 Ultra (60c)
Memory 256 GB
GPU Cores 60
Software
oMLX v0.3.8
macOS macOS 26.4.1
Context 32,768
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 1,186 64.6 47.8 GB view
4k 1,540 63.0 48.3 GB view
8k 1,586 59.6 48.5 GB view
16k 1,545 53.3 48.9 GB view
32k 1,416 40.1 50.5 GB current
64k 1,168 5.0 118.0 GB view
128k 853.1 13.6 120.8 GB view
Batching Results
Batch Size TG tok/s Speedup
64.6 1.00×
39.5 0.61×
70.1 1.09×