Prime Intellect Raises $130M for its Open Superintelligence Stack

Prime Intellect raises $130M in Series A funding led by Radical Ventures to scale its Open Superintelligence Stack for reinforcement learning and agents.

Prime Intellect closed a $130 million Series A round, led by Radical Ventures, with participation from NVIDIA Ventures, Intel Capital, Dell Technologies Capital, and existing investors, bringing its total funding to over $150 million. Several AI figures joined the round as angel investors, including John Schulman (Thinking Machines), Aaron Levie (Box), and Winston Weinberg (Harvey).

The company's thesis hinges on one idea: pre-training has concentrated frontier AI in a handful of labs, while reinforcement learning (RL) reshuffles the deck. It allows a company to own its optimization loop, train a model directly on its product, adapt it to its workflows, and deploy agents that continuously learn once in production. The necessary infrastructure has, until now, remained the exclusive domain of labs.

This is the stack Prime Intellect commercializes as the Open Superintelligence Stack: the same tooling used to train its own open models, made available to its clients. It covers compute, large-scale RL, environments, sandboxes, evaluations, and deployment, with dedicated or serverless serving and continual learning.

Going forward, the company targets long-horizon agents and Recursive Language Models (RLMs), models designed to manage their own context and coordinate sub-agents to handle multi-day tasks. It also aims for the automation of AI research and continual learning, where training and inference merge into a single loop.