The Jetson Orin Nano Super has been the entry-level Jetson since late 2024: 67 TOPS (INT8) for $249. NVIDIA raised the clock speeds and halved the price — the best entry point for local AI projects.
What you need
- Jetson Orin Nano Super Developer Kit
- MicroSD card (64 GB recommended, A2 class) or NVMe SSD
- USB-C power supply (at least 45 W — weak PSUs are the most common failure source!)
- Monitor + keyboard for initial setup (headless via SSH afterwards)
Step 1: Flash JetPack
The easiest options are the NVIDIA SDK Manager on a Linux host or the SD card image method:
- Download the JetPack image from the NVIDIA developer site
- Write it to the SD card with Balena Etcher
- Insert the card, connect power, boot — first-time setup (language, keyboard, user) runs in a dialog
Step 2: After first boot
sudo apt update && sudo apt upgrade -y
sudo nvpmodel -q # show current power model
jetson_release -v # check JetPack/CUDA versionStep 3: Measure performance
Don’t believe, measure: sudo jtop (from the jetson-stats package) shows clock speed, temperature, and power draw live.
pip3 install jetson-stats
jtopPractical experience: without a heatsink the Orin Nano throttles within minutes at full load. The included fan is fine for desktop loads; for sustained LLM inference I recommend active cooling.
Questions? Write me — webmaster@atlas.shrink2core.com or in the comments below.