The NVIDIA T4 is a power-efficient inference accelerator based on the Turing architecture, featuring 16GB of GDDR6 memory with 320 GB/s bandwidth in a low-profile PCIe form factor. With a 70W TDP, the T4 is optimized for high-throughput inference serving, video transcoding, and virtual desktop infrastructure. It introduced INT8 and INT4 Tensor Core precision for efficient inference and remains one of the most widely deployed inference GPUs in enterprise data centers and cloud environments.
Complete technical specifications for the T4.