NVIDIA Jetson is emerging as the standard platform for running open AI models at the edge, enabling real-time inference without cloud dependency. Use cases span industrial equipment (Caterpillar's in-cab AI assistant), dual-arm robotics (Franka FR3 Duo with GR00T N1.6), humanoid controllers (SONIC project at 50Hz), and personal AI assistants. Jetson Thor and Orin support a wide range of open models including Gemma 3, Mistral 3, Qwen 3.5, NVIDIA Cosmos, GR00T N1.6, Nemotron, and PI 0.5, with frameworks like vLLM, llama.cpp, Ollama, and SGLang. The platform consolidates compute and memory in a system-on-module, reducing hardware design complexity and sourcing challenges.
Table of contents
From the Cloud to the EdgeBuilding Autonomous Physical AI Systems at ScaleJetson Is the New StandardTake On Open Models on JetsonSort: