{"paper":{"title":"Magistral","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Adam Yang, Albert Q. Jiang, Alexander H. Liu, Alexandre Sablayrolles, Am\\'elie H\\'eliou, Am\\'elie Martin, Andy Ehrenberg, Andy Lo, Anmol Agarwal, Antoine Roux, Arthur Darcet, Arthur Mensch, Baptiste Bout, Baptiste Rozi\\`ere, Baudouin De Monicault, Chris Bamford, Christian Wallenwein, Christophe Renaudin, Cl\\'emence Lanfranchi, Darius Dabert, Devon Mizelle, Diego de las Casas, Elliot Chane-Sane, Emilien Fugier, Emma Bou Hanna, Gabrielle Berrada, Gauthier Delerce, Gauthier Guinet, Georgii Novikov, Guillaume Lample, Guillaume Martin, Himanshu Jaju, Jan Ludziejewski, Jason Rute, Jean-Hadrien Chabran, Jean-Malo Delignon, Joachim Studnia, Joep Barmentlo, Jonas Amar, Josselin Somerville Roberts, Julien Denize, Karan Saxena, Karmesh Yadav, Kartik Khandelwal, Khyathi Raghavi Chandu, Kush Jain, L\\'elio Renard Lavaud, L\\'eonard Blier, Lingxiao Zhao, Louis Martin, Lucile Saulnier, Luyu Gao, Marie Pellat, Mathilde Guillaumin, Mathis Felardos, Matthieu Dinot, Maxime Darrin, Maximilian Augustin, Micka\\\"el Seznec, Mistral-AI: Abhinav Rastogi, Neha Gupta, Nikhil Raghuraman, Olivier Duchenne, Patricia Wang, Patrick von Platen, Patryk Saffer, Paula Kurylowicz, Paul Jacob, Paul Wambergue, Pavankumar Reddy Muddireddy, Philom\\`ene Chagniot, Pierre Stock, Pravesh Agrawal, R\\'emi Delacourt, Romain Sauvestre, Roman Soletskyi, Sagar Vaze, Sanchit Gandhi, Sandeep Subramanian, Shashwat Dalal, Siddharth Gandhi, Soham Ghosh, Srijan Mishra, Sumukh Aithal, Szymon Antoniak, Teven Le Scao, Thibault Schueller, Thibaut Lavril, Thomas Robert, Thomas Wang, Timoth\\'ee Lacroix, Valeriia Nemychnikova, Victor Paltz, Virgile Richard, Wen-Ding Li, William Marshall, Xuanyu Zhang, Yihan Wang, Yunhao Tang","submitted_at":"2025-06-12T17:22:37Z","abstract_excerpt":"We introduce Magistral, Mistral's first reasoning model and our own scalable reinforcement learning (RL) pipeline. Instead of relying on existing implementations and RL traces distilled from prior models, we follow a ground up approach, relying solely on our own models and infrastructure. Notably, we demonstrate a stack that enabled us to explore the limits of pure RL training of LLMs, present a simple method to force the reasoning language of the model, and show that RL on text data alone maintains most of the initial checkpoint's capabilities. We find that RL on text maintains or improves mu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.10910","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2506.10910/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}