There Are A Lot Of Dark Matter Models Out There

There are more than six thousand dark matter models floating out there in theory space. This is a lot, but it is also eminently finite. So, it makes sense that AI is a good tool to use to start making sense of them. But AI is not terribly accurate (81%), even though, in this case, it didn't actually hallucinate any new models.

Of course, there is only one right answer (which isn't guaranteed to be any of the models proposed so far), and a large share of them are ruled out be existing evidence (of which the creators of many models weren't necessarily aware).

To use a dark-matter model one needs to know what defines it, which observables it affects, which codes and likelihoods evaluate it, and under what assumptions a published bound applies. That information is scattered across tens of thousands of papers. 

We present DarkMatterFM-Registry, a machine-readable registry of dark-matter models extracted from arXiv full texts by a staged large-language-model (LLM) pipeline: 6,145 structured model cards, validated against a seven-file JSON Schema suite, ranked by expected value for observational follow-up, and enriched for the top ~2,000 with observables, constraints, and theory-to-observable routes. 

Schema checks pass 97.8% of cards, and an LLM-judged accuracy study on a stratified card sample finds 81% of section verdicts correct, 18% partially correct, and none hallucinated. 

As a first application we turn the registry into a priority-weighted roadmap of 25 concrete projects the community could build. The roadmap is built by judging every computational step of every enriched model's theory-to-observable routes as available, partial, or missing, and aggregating the resulting typed gaps weighted by model priority. 

We find that what blocks the most high-priority models is not missing theory but missing shared infrastructure: model files for existing relic-abundance solvers, a general non-cold Boltzmann-code framework, reusable indirect-detection likelihoods, and a Lyman-α forest likelihood for non-cold dark matter (the largest gap for structure-formation probes). The model cards, schemas, roadmap, and ranking are released making the project reproducible and the cards available for the community.

Ioana A. Zelko, Jo Bovy, Nolan Koblischke, "DarkMatterFM-Registry: A Machine-Readable Registry of Dark-Matter Models Extracted from the arXiv Literature" arXiv:2611.08583 (November 5, 2026)  (Accepted at the Machine Learning and the Physical Sciences Workshop (ML4PS), NeurIPS 2026. Registry, schemas, and code: this https URL).