Patronus AI announced a US$50 million Series B led by Greenfield Partners and previewed Digital World Models for simulating digital environments used to train and evaluate agents, the company said on 25 June 2026.
AI evaluation vendors are raising capital to simulate agent environments beyond static prompt suites. Buyers still need evidence that world-model evals catch their production failure modes.
Patronus AI sells evaluation and oversight for LLM applications. Lightspeed, Notable Capital, Datadog, Samsung and others participated per company materials.
ML platform owners should demand eval coverage on their critical journeys — before treating any Series B world-model preview as a completed AI quality programme.