NVIDIA Jetson Models

NVIDIA Jetson Model Comparison

Here's a snapshot of the major Jetson modules, from entry-level to cutting-edge:

Model	AI Performance	Memory	Power Usage	Target Use Case
Jetson Nano	~0.5 TOPS	4 GB	~5–10 W	Basic AI, IoT, hobbyist projects
Jetson TX2	~1.3 TOPS	8 GB	~7.5–15 W	Industrial edge AI, drones
Jetson Xavier NX	~21 TOPS	8–16 GB	~10–15 W	Smart cameras, compact robotics
Jetson AGX Xavier	~32 TOPS	32–64 GB	~15–30 W	Autonomous machines, advanced robotics
Jetson Orin Nano	Up to 67 TOPS	4–8 GB	~7–25 W	Entry-level edge AI with high efficiency
Jetson Orin NX	Up to 157 TOPS	8–16 GB	~10–40 W	Mid-range robotics, multi-sensor systems
Jetson AGX Orin	Up to 275 TOPS	32–64 GB	~15–60 W	High-performance robotics, AI pipelines
Jetson AGX Thor	Up to 2070 TFLOPS (FP4)	128 GB	~40–130 W	Humanoid robots, generative AI at edge

The Jetson ecosystem is booming, especially in these areas:

The newly launched Jetson AGX Thor is the most advanced yet, powering real-time reasoning and generative AI models directly on-device.
Companies like Boston Dynamics, Amazon Robotics, and Agility Robotics are already integrating Thor into their next-gen robots.

Jetson Orin NX and AGX Orin are widely adopted for smart factories, logistics, and autonomous machines due to their balance of power and efficiency.

Jetson Orin Nano is becoming the go-to for compact, low-power deployments with surprisingly high AI throughput—ideal for smart cameras, drones, and mobile robots.

Jetson Nano remains popular for learning and prototyping, though its performance is now dwarfed by newer models.

Unified software stack: JetPack SDK and CUDA-X make development seamless across models.
Scalability: Developers can prototype on Nano or Xavier NX and scale up to Orin or Thor without rewriting code.
Ecosystem: Over 2 million developers and 150+ partners are building on Jetson.