NVIDIA 900-13834-0000-001 | Jetson Thor T4000 Module, 1200 FP4 TFLOPS, 64GB

Q: What is the NVIDIA Jetson T4000 Module designed for?
A: The T4000 module is designed for physical AI, robotics, and edge inference applications including autonomous mobile robots, quadrupeds, drones, humanoid robotics, industrial automation, and visual AI systems requiring real-time multimodal sensor processing and generative AI model execution.

Q: How does the T4000 compare to the Jetson AGX Orin?
A: The T4000 delivers approximately 4x higher AI compute performance than the Jetson AGX Orin with exceptional energy efficiency, powered by the newer Blackwell GPU architecture versus Orin's Ampere architecture, while maintaining a compact power envelope of 40W to 70W.

Q: Is the T4000 pin-compatible with other Jetson Thor modules?
A: Yes, the T4000 shares form factor and pin compatibility with the higher-performance T5000 module, allowing developers to scale between performance tiers without redesigning carrier boards, simplifying hardware development and deployment flexibility.

Q: What AI models and frameworks does the T4000 support?
A: The T4000 supports a wide range of generative AI models including Vision Language Action (VLA) models like NVIDIA Isaac GR00T for humanoids, popular large language models (LLMs), vision language models (VLMs) such as Cosmos Reason, and runs the full NVIDIA AI software stack including JetPack 7, CUDA 13.0, TensorRT, Isaac, Metropolis, and Holoscan.

Q: What makes the Blackwell architecture significant for this module?
A: The Blackwell GPU architecture introduces native FP4 quantization with a next-generation Transformer Engine and Multi-Instance GPU (MIG) technology, enabling predictable performance in mixed-criticality robotics applications while efficiently running large transformer models and generative AI workloads at the edge with lower power consumption.