
Prime Intellect
Prime Intellect trains and serves frontier open models on Nebius. Its SYNTHETIC-2 and INTELLECT-2 runs used Nebius GPU clusters, including multi-region distributed training across Nebius data centers in Europe and the US and a GB200 NVL72 proof of concept.
Nebius is also a named inference provider for INTELLECT-3. The collaboration demonstrates decentralized and distributed training at scale on Nebius AI Cloud infrastructure.
See Nebius case study: Prime Intellect: Distributed training and RL with NVIDIA GB200 NVL72.