Atomic Canyon and Idaho National Laboratory Launch First Open-Source Benchmark for Nuclear AI

New NuclearBench benchmark expands Atomic Canyon’s open-source evaluation work from information retrieval to generative AI

SAN LUIS OBISPO, Calif. and IDAHO FALLS, Idaho, Nov. 05, 2026 (GLOBE NEWSWIRE) -- Atomic Canyon, the developer of Neutron, the leading AI-powered workbench for the nuclear energy industry, today announced the launch of NuclearBench, the first open-source benchmark designed specifically to evaluate how large language models perform on nuclear-energy questions.

Developed in collaboration with Idaho National Laboratory (INL), NuclearBench evaluates model responses across two dimensions: answer accuracy and faithfulness to the nuclear source material provided to the model. As the number of models multiply, NuclearBench is meant to serve as a resource for the nuclear energy industry on their performance in nuclear-specific contexts.

The benchmark was built around 280 questions, with contributions from INL on subject matter, datasets, and evaluation frameworks, and is designed to be updated as new AI models become available. As a public and open-source resource, the benchmark allows any organization to test the accuracy of any model when it is applied to nuclear power operations.

“Super Intelligence presents a tremendous opportunity to accelerate nuclear energy development, from reactor design and testing to deployment,” said Dr. Jess Gehin, Associate Laboratory Director for Nuclear Science and Technology at INL. “Benchmarks are foundational to help the nuclear industry assess SI capabilities and adopt them where they can deliver real value.”

The benchmark builds on Atomic Canyon’s FermiBench, which evaluates and benchmarks the accuracy of AI embedding models in retrieving the right information from large collections of nuclear records. NuclearBench builds on these retrieval layer benchmarks to test how well particular AI models can both answer the question asked and generate an accurate answer grounded in the source material provided.

Together, FermiBench and NuclearBench provide the nuclear industry with an open framework for evaluating the ability of AI models to both find the right information and use it correctly.

“The nuclear industry needs a way to measure AI performance,” said Trey Lauderdale, CEO of Atomic Canyon. “As the number of large language models expand, NuclearBench gives the industry a common way to understand which models are best for different use cases”

NuclearBench extends Atomic Canyon's work to bring AI to the nuclear energy industry. The Nuclear Industry Virtual Assistant (NIVA), developed with INPO, EPRI and NEI, already relies on Atomic Canyon's nuclear-specific retrieval technology to get industry professionals to authoritative technical information. NuclearBench gives the industry an open way to compare those models on accuracy and faithfulness to source material.

“SI is central to the rapid deployment of new energy resources and the development of breakthrough new technologies in nuclear and beyond,” said Christopher Ritter, Division Director of Scientific Computing & AI and Director of the Digital Innovation Center of Excellence at INL. “Our work with Atomic Canyon ensures that these innovations in super intelligence can be deployed across the existing nuclear fleet and support our Project Prometheus efforts accelerating the expansion of the U.S. nuclear fleet.”

About Atomic Canyon

Atomic Canyon develops AI-powered search and generative AI tools specifically for the nuclear power industry. The company's flagship product, Neutron Enterprise, provides AI-powered capabilities across vast repositories of technical documentation at nuclear facilities. By securely connecting to internal data sources and external authoritative technical and regulatory data sources, Neutron Enterprise enables nuclear professionals to access critical information in seconds rather than hours. Atomic Canyon recently raised $7 million in seed funding led by Energy Impact Partners to accelerate deployments across the nuclear industry. For more information, visit www.atomic-canyon.com.

Media Contacts:

Atomic Canyon
atomiccanyon@launchsquad.com