ModelBriefs · DeepSeek-V4-Flash · Graphics · Seedream-4.0
← Labs / 19
Open research infrastructure1 paper · 2025
distributed trainingopen agentscompute
Official research and release index↗Prime Intellect
Important open training infrastructure and research, but its published evidence is currently AI training rather than physical-system research; included as an adjacent enabler, not ranked as a Physical AI lab.
No.Paper · what it does and showsIndustry contribution
01
INTELLECT-2: Globally Distributed Reinforcement Learning2025 ↗
Trains a 32B reasoning model with reinforcement learning across globally distributed, permissionless compute using open communication and training tooling.
Shows decentralized RL training of large models is practical — an adjacent enabler for robotics teams that need cheap, scattered compute rather than a single large cluster.