The U.S. Department of Energy has announced a new class of open-weight foundation models designed to accelerate scientific discovery, built under its Genesis Mission initiative. The first model is Genesis-Science-1 (GS1), developed in partnership with Arcee AI. Alongside the announcement, a DOE-hosted contribution portal — run by Argonne National Laboratory — went live, inviting the research community to contribute data, tasks, and evaluations, with first-round applications due August 14, 2026.
A national government directly funding open-weight foundation models is rare. The signal matters more than any single model.
Genesis-Science-1: The First Model, Led by Arcee
The division of labor is clear. Arcee AI leads model development: securing compute, curating training data, running pretraining and post-training, building the governed execution environment, and handling evaluation and release. DOE and participating national laboratories keep the system anchored in real scientific practice — supplying reviewed scientific materials, defining representative research tasks, designing evaluations, and validating the system’s results.
Arcee describes GS1 as an open-weight model paired with a governed research system: the goal is not answering questions but completing scientific computing workflows end to end while preserving a reproducible record of the work. Arcee CEO Mark McQuade put the stakes bluntly: “A country cannot lead in AI if everything it leads in is closed.” He added that GS1 is being held to a demanding standard — useful scientific work under real operating constraints, judged by the people who actually do the science.
A National Contribution Portal: Data, Tasks, Evaluations
The portal is open to universities, national laboratories, companies, scientific nonprofits, and research organizations, with two contribution tracks scheduled for 2026. Foundation-stage contributions target material suitable for pretraining, midtraining, or context extension: scientific text, code, documentation, and structured technical collections. Post-training contributions get more specific: expert demonstrations, annotated task examples, research software, datasets, complete workflow environments, reinforcement-learning tasks, held-out evaluations, scoring rubrics, tests, and verifiers. The program is also recruiting scientists and engineers to review tasks, model behavior, and completed artifacts.
The application process itself is deliberately lightweight: the public form collects descriptions and metadata only, and no scientific material is transferred during application. Selection runs through five gates — scientific fit, rights and handling, expert and evaluation readiness, technical integration, and final program selection. Accepted contributors get early evaluation access as the system develops and credit in the technical report and release materials.
What stands out is how concrete the scope is for a government program. Rather than asking for generic “AI data,” the portal asks scientists for the artifacts of real work: experimental and observational data from user facilities, simulation outputs and run logs from supercomputing campaigns, materials and chemistry collections, and the tools researchers use to conduct and review their work. Materials only enter after completing the Department’s own release-review process, so provenance is handled up front rather than litigated later. The August 14 deadline applies to the first contribution window; infrastructure providers can submit rolling inquiries about training and evaluation capacity.
From Executive Order to National Open Models
Genesis Mission was launched by executive order in November 2025 as a Department of Energy-led initiative, with a stated goal of doubling the productivity and impact of American science and engineering within a decade. It connects DOE’s 17 national laboratories, leadership-class computing facilities, and partners across industry and academia, under the direction of Under Secretary for Science Darío Gil. GS1’s target domains span materials discovery, energy systems, earth systems modeling, fusion, biology, and high-energy physics. The official framing rests on three principles — open science, reproducibility, and responsible AI — with the models positioned as a shared public asset.
What It Means for Open Source and Science
Two implications stand out. For the open-source ecosystem, this is a major government treating open weights as an instrument of national science and technology strategy — a notable marker in the ongoing policy debate over American open models, and one that Arcee’s quote summarizes neatly. For the research community, it is a concrete opportunity: experimental data, simulation outputs, and research software that rarely become model training material now have an official path into a foundation model, with governance terms declared before contribution rather than negotiated after release. And DOE’s strict release-review process means this will not be a “dump the weights online” kind of open source. Whether enough institutions actually apply before the August 14 deadline is the first thing to watch.
Sources
- Genesis Open Models — U.S. Department of Energy
- The Genesis Mission — Department of Energy
- Announcing Genesis-Science-1 — Arcee AI
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.