The difference is that a neural network you can fit on a drone is going to be a lot less capable than an LLM you can run on a desktop.
You can fit a Jetson Orin Nano on a winged drone. It has plenty enough power to run a vision model.
Doesn't a fiber tether give its drone desktop-class computing?
You can fit a Jetson Orin Nano on a winged drone. It has plenty enough power to run a vision model.