SAN LUIS OBISPO, Calif., and IDAHO FALLS, Idaho — September 9, 2025 — Leads & Copy — Atomic Canyon and Idaho National Laboratory (INL) are partnering to develop and release a benchmark suite for evaluating retrieval-augmented generation (RAG) and large language models (LLMs) in nuclear applications. The partnership aims to establish standards for AI adoption in nuclear facility operations, addressing the need for objective evaluation methods of generative AI systems in nuclear environments.
The benchmark suite will allow nuclear operators to assess AI systems that access public nuclear documentation through retrieval-augmented generation. INL will contribute nuclear expertise and access to public datasets, focusing on real-world nuclear workflows like document retrieval and regulatory compliance checks.
Atomic Canyon’s CEO, Trey Lauderdale, emphasized the importance of the collaboration, stating it will establish a gold standard for evaluating AI systems in nuclear environments. Christopher Ritter, Division Director at Idaho National Laboratory, said the partnership would create the benchmarks needed to deploy AI across the nuclear enterprise.
The project will curate datasets, define tasks, and produce documentation over six months, releasing all data under open-source licenses. Atomic Canyon will develop and release accompanying evaluation software and tools under permissive open-source licenses.
Media Contacts:
Atomic Canyon: atomiccanyon@launchsquad.com
Idaho National Laboratory: Sarah.Neumann@inl.gov
Source: Atomic Canyon
