CERN and Mistral have signed a statement of intent to establish a long term strategic partnership focused on foundation models for scientific discovery. The centrepiece is a single pretrained Large Physics Model, which CERN describes as the first foundation model for collider physics. It will be trained on billions of simulated proton proton collisions generated at the Large Hadron Collider. CERN says the work pairs its expertise in fundamental physics, scientific data and large scale computing with Mistral's strength in training large models, with an initial focus on high energy physics.
The model targets a long standing bottleneck in LHC analysis. Reconstruction today runs through a chain of specialised algorithms that progressively convert raw detector signals into the particles and objects used for physics analysis. Machine learning already improves individual steps, but a foundation model could learn across the entire reconstruction chain instead. Planned tasks include reconstructing collisions, identifying particles, flagging unexpected signals, and evaluating future detector designs. The scale is enormous: CERN has stored roughly one exabyte of LHC experimental data, and collision numbers are set to rise tenfold when the High Luminosity LHC starts operating around 2030.
Both sides framed the deal as a European pairing of AI capability with fundamental science. Enrica Porcari, CERN's Chief Information Officer, said high energy physics offers a unique environment where AI can be developed and tested with exceptional rigour, and that the aim is capabilities the whole physics community can share. Johannes Brandstetter, Mistral's Vice President of AI for Engineering, said the partnership gives the two organisations a chance to push the frontiers of particle physics. While the first focus is collider physics, the partners expect the methods developed here to serve the broader scientific community.