CERN and Mistral signed a statement of intent for a long-term partnership. They aim to develop foundation models for scientific discovery. The first goal is a Large Physics Model trained on LHC collision data. It will handle reconstruction, particle identification, signal flagging and detector evaluation. Today’s process uses separate algorithms for each step. The model could extend beyond high-energy physics to the broader scientific community. CERN stores one exabyte of LHC data. HL-LHC will increase collision volume tenfold around 2030. Enrica Porcari called the partnership a chance to advance AI for physics. Johannes Brandstetter said it could push high-energy physics frontiers.
What changed
Reconstruction relied on a chain of specialised algorithms for each task.
What it unlocks
Using one model for reconstructing collisions, identifying particles, flagging signals and evaluating detector designs.
- 1 exabyte of LHC data stored
- 10x more collisions with HL-LHC
- 2030 HL-LHC start date
What you need to act on it
- a long-term strategic partnership between CERN and Mistral
Sources