← Back to community benchmarks

Ternary-Bonsai-27B

M3 Ultra (60c) · 256 GB · 2bit · 2026-07-20
Performance
16k
tokens
331.5
PP tok/s
42.2
TG tok/s
49421
TTFT (ms)
10.9
Peak mem (GB)
Hardware
Chip M3 Ultra (60c)
Memory 256 GB
GPU Cores 60
Software
oMLX v0.5.1
macOS macOS 26.5.2
Context 16,384
Performance by Context Length
Context PP tok/s TG tok/s Peak Mem
1k 350.6 48.2 8.9 GB view
4k 354.1 46.9 10.1 GB view
8k 346.4 45.1 10.6 GB view
16k 331.5 42.2 10.9 GB current
32k 306.4 37.0 12.1 GB view
64k 266.2 32.7 15.0 GB view
Batching Results
Batch Size TG tok/s Speedup
48.2 1.00×
73.5 1.52×
93.0 1.93×
105.9 2.20×