27B Model Tested as Usable on Jetson
BoogerheadCult · reddit · 2026-07-15
PrismML Bonsai 27B is usable on the Jetson Orin Nano 8GB. The author shared a set of local benchmark and resource metrics:
- Context length: 48k
- Prompt processing speed: 27 tok/s
- Generation speed: 4.31 tok/s
- Memory usage after cold start load: 6.2GB (no-mmap)
- Power consumption: 25W
The author concludes that while the speed isn't exceptionally fast, this low-power edge AI configuration handles complex tasks quite well.
More from Infra
- Nearly half of Claude revenue now flows through cloud marketplace commitments, up from 11% — BenBajarin · 2026-10-26
- Anthropic ARR goes vertical after price cuts: tokens up 4x, $100B+ ARR back in play — GavinSBaker · 2026-10-26
- Alex Banks argues power caps GPU supply, real risk is demand cooling first — thealexbanks · 2026-10-26
- Analyst: Ant has the most fungible AI workload of any model provider today — BenBajarin · 2026-10-26
- Anthropic S-1: training infra edge cut per-task cost 40% and doubled revenue growth — BenBajarin · 2026-10-26
- Rewriting scaling law study: optimal synthetic data share rises from 10% to 32% with compute — heghbalz · 2026-10-26