The people behind the CERN and Mistral partnership
Two people are quoted in the announcement. Here is who they are, and the researchers whose work the model follows.
The announcement names two people. Neither CERN nor Mistral has said who will lead the work day to day or how big the team is. We add names here only when one of them publishes them.
Enrica Porcari, CERN
CERN's Chief Information Officer, quoted in the release and by Reuters. She described high-energy physics as a place where AI can be tested rigorously, and said the partnership should produce AI that serves the whole physics community and can be shared (CERN, Reuters).
Johannes Brandstetter, Mistral
Named in CERN's release as Mistral's Vice-President of AI for Engineering. An Austrian investment agency's report of the same year gives his title as Vice President, AI for Science (Invest in Austria), so the title may have changed. He co-founded Emmi AI, a physics-AI company spun out of JKU Linz in Austria, and joined Mistral when it agreed to buy Emmi in May 2026 (Mistral). Before that he was one of the authors of Aurora, Microsoft's foundation model of the atmosphere, published in Nature in 2025 (arXiv 2405.13063). So he has helped build a large science foundation model before.
Arthur Mensch, Mistral
Mistral's co-founder and chief executive. He wasn't quoted in the release, but wrote in Le Monde on the day of the announcement that frontier models are now indispensable to science (as reported by Reuters).
The researchers whose work it follows
None of these people are named in the announcement. Their published work is what a collider foundation model builds on:
- Joosep Pata and colleagues in CMS, who built machine-learned particle flow, a single model that does much of the reconstruction work (CERN).
- Gregor Kasieczka, Anna Hallin and Joschka Birk at the University of Hamburg (OmniJet-α); Hallin also wrote the first review of the field (arXiv 2509.21434).
- Vinicius Mikuni and Benjamin Nachman at Berkeley Lab (OmniLearn, OmniLearned and the anomaly-search work).
- Tobias Golling, Lukas Heinrich and Michael Kagan (Masked Particle Modeling).