Just days ago, billionaire Elon Musk unveiled Grok-3, the latest AI model from his company xAI, boasting that it was "the most intelligent artificial intelligence on Earth." Of course, Musk wasted no time showcasing his AI’s capabilities, even though Grok has embarrassed him more than once in the past.

Most recently, Musk shared a tweet from user Luis Batalha, praising Grok-3’s “extraordinary” progress. The tweet claimed that Grok-3 had solved the hardest problem in the 2025 Putnam Mathematics Competition.

---

Grok-3 vs. the World’s Hardest Math Exam

The Putnam Mathematics Competition is an annual exam for undergraduate students in the U.S. and Canada, previously recognized by Time Magazine as the most difficult math competition in the world.

Statistics show that in the 2024 Putnam exam, which lasts six hours, the highest score achieved was only 90 out of 120, while the average score among nearly 4,000 participants was just 8 points.

In the 2025 Putnam Competition, over 500 competitors failed to solve a specific matrix determinant problem under the given conditions.

Yet, Grok-3’s “Think Mode” allegedly solved it in just 8 minutes.

AI’s Victory Short-Lived: Grok-3 Admits Its Own Mistake

Musk’s excitement was short-lived as an unexpected twist unfolded.

After Grok-3’s solution was published, many mathematicians, programmers, and curious users decided to double-check the answer—using Grok-3 itself.

The result was shocking: Grok-3 admitted that its previous solution was incorrect!

Users shared screenshots of conversations with the AI, in which Grok-3 acknowledged a logical flaw in its reasoning when verifying the Putnam problem’s requirements.

Not the First Time Grok-3 Has Embarrassed Musk

This isn’t the first time Grok-3 has left Musk red-faced.

Previously, screenshots circulated showing the AI accusing Musk—its own creator—of spreading misinformation. Not long after, Grok’s development team had to disable its ability to answer questions related to “sources proving Musk spreads fake news.”

This latest incident raises serious questions about AI reliability, even as tech leaders like Musk continue to push for more advanced AI models. Whether Grok-3’s mistake was an anomaly or a sign of deeper flaws, it’s another humbling moment for Musk and xAI.