PSI's Turturean solves 18-year-old quantum problem with Fable 5.2

The Aaronson-Ambainis conjecture had stood for 18 years as one of the last centrally important open problems in quantum query complexity, so its resolution by a recent MIT graduate working at a startup is a notable shift in who produces foundational results. Scott Aaronson's reaction — surprised it was missing from OpenAI's recent math-focused release — points to a gap between broad model capabilities and targeted research systems.

PSI's approach is to wrap models in what CEO Matt Pines calls harnesses, tools, memory systems and simulation infrastructure, letting the system attack a problem at breadth and depth rather than relying on a single model pass. That framing helps explain why a startup rather than a major lab got there first, and why Pines says theoretical physics is not yet mostly the domain of machines but is on its way.

The company's commercial base is narrower than its ambitions: revenue today comes from optimizing data centers by simulating heat, power, water and sound. Pines acknowledges that physics differs from math because nature gets a vote, and PSI is focused on validating computational results through experiments. The academic side is still catching up on norms — PSI held its result until Anthropic's model was nearly ready for release, shared it with researchers, and Halverson will join an MIT summit next week where physicists try to agree on a response.