Join NVIDIA at NeurIPS 2025

La-Proteina is the first generative model that demonstrates accurate codesign of fully atomistic protein structures at scale, up to 800 residues, with state-of-the-art atomistic motif scaffolding performance. 

Kinetic GROVER Multi-Task (KERMT), an enhanced reimplementation of the GROVER model, uses PyTorch Distributed Data Parallel for distributed pretraining, automates hyperparameter tuning, and accelerates fine-tuning and prediction using cuik-molmaker.

nvMolKit is a GPU-accelerated cheminformatics library developed and designed to significantly speed up molecular computation tasks common in computational chemistry and drug discovery.

When combined with MMseqs2-GPU on an x86 system with one NVIDIA RTX PRO™ 6000 Blackwell Server Edition GPU, deep learning inference with OpenFold and TensorRT™ is up to 131x faster.

Learn how to transform sequential optimization algorithms into parallel ones that can leverage GPU acceleration to efficiently manage complex, high-dimensional investment portfolios.

See how Numba on NVIDIA H200 Tensor Core GPUs increased performance of computational simulations of financial markets by up to 114x compared to CPU-only simulations.

Discover how Nasdaq is accelerating financial insights and boosting AI performance with NVIDIA NeMo™ Retriever and NIM™. Learn how scalable AI improves response time, accuracy, and cost efficiency.

Learn how developers in capital markets can streamline trade capture and evaluation with self-correcting AI workflows with NVIDIA NIM.