Plate Nº 86 · recorded October 16, 2026
PhysicsReported finding
CERN and Mistral to Build First Foundation Model for Collider Physics
CERN and Mistral will develop a Large Physics Model trained on billions of simulated LHC collisions, aiming to unify reconstruction tasks now handled by separate specialised algorithms.
By Priya Raman3 min read641 words
In brief
- CERN and Mistral signed a statement of intent on 15 October 2026 to develop AI foundation models for science.
- The planned Large Physics Model would be the first foundation model for collider physics, trained on billions of simulated LHC collisions.
- CERN stores around one exabyte — one million terabytes — of LHC experimental data.
- Collision rates are set to grow tenfold with the HL-LHC, expected to begin operation around 2030.
On 15 October 2026, CERN and the French AI company Mistral signed a statement of intent to develop what they describe as the first foundation model for collider physics — a single pretrained system intended to handle tasks across the entire analysis chain of the Large Hadron Collider (LHC), the world's largest and most powerful particle accelerator.
CERN, the European Laboratory for Particle Physics, brings to the partnership roughly one exabyte of stored experimental data — one million terabytes — collected from the LHC to date. Mistral, a European developer of large-scale AI models, brings its expertise in training such systems. The long-term strategic partnership will focus first on high-energy physics, though the two organisations say the AI approaches it produces could serve the broader scientific community.
What is a "Large Physics Model"?
A foundation model (an AI system pretrained on vast datasets that can then perform many different tasks) is the organising idea behind the collaboration. The planned "Large Physics Model" will train on billions of simulated proton-proton collisions from the LHC.
Its designers intend it to support multiple jobs at once:
- reconstructing particle collisions from raw detector signals;
- identifying which particles were produced;
- flagging unexpected signals that might point to new physics;
- evaluating designs for future detectors.
Why does CERN need it?
Today, reconstructing an LHC collision relies on a chain of specialised algorithms that progressively transform detector signals into the particles and other objects physicists analyse. Machine learning has long played a role in this process, improving performance on individual tasks.
Foundation models offer a different route: instead of tuning separate tools for each step, a single model could learn across the whole reconstruction chain and support multiple tasks at once. That matters because the volume of data is about to grow sharply. The High-Luminosity LHC (HL-LHC), an upgraded phase of the collider expected to begin operation around 2030, will increase the number of collisions tenfold.
The scale and precision of CERN's data make the laboratory, in the partners' framing, an unusually rigorous testing ground for AI.
"High-energy physics provides a unique environment in which artificial intelligence can be developed and tested with exceptional rigour," said Enrica Porcari, CERN's Chief Information Officer. "We are partnering with Mistral, a leading European AI company, to advance AI capabilities that can serve the whole physics community and be shared."
Mistral, headquartered in France, builds open-weight models — AI systems whose trained parameters are publicly available for others to download, customise and run. The company maintains an international presence across Europe, North America and Asia.
How ambitious is the timeline?
The document signed is a statement of intent rather than a finished research programme, and neither organisation has announced a delivery date for the Large Physics Model or specified its architecture. The announcement also leaves open how the trained model would be shared beyond the physics community, beyond Porcari's stated intention that the capabilities "be shared".
Johannes Brandstetter, Vice-President of AI for Engineering at Mistral, framed the collaboration in exploratory terms. "We are proud to work with CERN to explore what large-scale AI models can bring to the growing complexity of particle physics," he said. "Together, we have the opportunity to push the frontiers of high-energy physics, developing AI capabilities that have the potential to lead to new scientific discoveries."
Why it matters for Europe
Beyond the physics, the partnership carries a strategic signal. By pairing European excellence in fundamental science with a leading European AI developer, CERN and Mistral present the collaboration as a step toward a new generation of AI tools capable of matching the scale, precision and complexity of modern research — in Europe and beyond.
For now, the work begins with simulated collisions, one exabyte of archived data, and a tenfold increase on the horizon.
By Priya Raman
via mistral.ai (Original)
More from Priya Raman
Show full bio
Senior reporter covering people, launches and day-to-day practice in general science news at SciBeat.
242 articles
Nearby plates
- NSF Names Three X-Labs Topics, Opens AI-for-Physical-Systems Call
- CERN's Super Proton Synchrotron Marks 50 Years of Particle Physics Firsts
- NSF commits $90 million to three new science and technology centers
- CERN Begins Replacing 28 LHC Magnets in HiLumi Overhaul
- First ITk Petals Installed in ATLAS Strips Endcap at DESY