Source · Tier B · Technical report

INTELLECT-2: A Reasoning Model Trained Through Globally Decentralized Reinforcement Learning

Prime Intellect Team, S. Jaghouar, J. Mattern, J. M. Ong, J. Straube, M. Basra, A. Pazdera, K. Thaman, M. Di Ferrante, F. Gabriel, F. Obeid, K. Erdem, M. Keiblinger, J. Hagemann. 2025. arXiv.

Linkhttps://arxiv.org/abs/2505.07291
arXiv2505.07291
VersionRead arXiv v1 (HTML), section 2.3 on TOPLOC and section 2.4 on validation.
Accessed2026-09-23
OrganizationPrime Intellect
NoteDeveloper's report on its own training run; TOPLOC statements are provider-reported.

Cited by