Based on NVIDIA's official announcement, at the SC25 conference, NVIDIA formally introduced the NVIDIA Apollo series. This is an open - source family of AI physical models designed to expedite industrial and computational engineering processes. Alongside this, NVIDIA also launched the Warp open - source Python framework tailored for physical simulation.

The Apollo series models incorporate cutting - edge architectures, such as neural operators, Transformer models, and diffusion methods. These advanced features render them well - suited for a diverse range of fields, including the automation of electronic devices, the semiconductor industry, structural mechanics, as well as weather and climate studies.

The Warp framework seamlessly blends the user - friendliness of Python with the high - performance capabilities at the CUDA level. It can achieve GPU acceleration up to 245 times faster and enables the development of GPU - accelerated 3D simulation workflows.