Argonne Opens AI Inference Service for Department of Energy Researchers
Argonne National Laboratory launched a cloud-based AI inference service on Tuesday, giving Department of Energy researchers remote access to advanced AI models for scientific work. The platform runs on some of the nation's most powerful computing systems, including the exascale computer Aurora and NVIDIA's DGX A100 cluster.
The service addresses a practical problem: scientists spend weeks analyzing data when they could spend hours. By offering pre-trained models as a shared resource, researchers can test hypotheses faster without building and maintaining their own infrastructure.
What's Available
The inference service provides large language models and domain-specific science foundation models. Current offerings include open-weight models and models developed in-house at Argonne.
The hardware runs across three systems: Aurora, the NVIDIA DGX A100 cluster called Sophia, and the SambaNova SN40L chip cluster named Metis.
Who Can Access It
Eight national laboratories can use the service:
- Los Alamos National Laboratory
- Brookhaven National Laboratory
- Lawrence Berkeley National Laboratory
- Fermi National Accelerator Laboratory
- Lawrence Livermore National Laboratory
- Oak Ridge National Laboratory
- Sandia National Laboratories
- Thomas Jefferson National Accelerator Facility
Broader Applications
While the service supports the Department of Energy's Genesis Mission-a program to advance AI research using federal datasets and resources-it extends beyond AI research. The platform can apply to fusion energy, chemistry, and materials science projects.
Michael Papka, director of the Argonne Leadership Computing Facility, said the service "helps close the gap between developing AI models and putting them to work in scientific research." Venkat Vishwanath, AI and machine learning lead at the facility, added that researchers can now "rapidly interpret results, refine experiments and explore complex systems in ways that weren't practical before."
The infrastructure builds on a 2025 framework designed to let researchers run multiple AI tasks in parallel on different models without relying on commercial cloud services.
For researchers looking to develop these skills, AI for Science & Research courses cover practical applications in laboratory settings and data analysis.
Your membership also unlocks:
