The Office of Justin Sun has announced the first confirmed recipients of the Justin Sun Prize: Wouter van Doorn, Quanyu Tang, and Yanyang Li. The three were recognized for contributions to six Erdős problems, the mathematical questions posed or popularized by Hungarian mathematician Paul Erdős. It's the program's first batch of awards, and it covers both original discovery and the unglamorous work of making proofs machine-checkable.
A prize built around proof, not prestige
The Justin Sun Prize is an academic initiative established by Justin Sun, the founder of TRON and Grenada's former ambassador to the World Trade Organization. The program is decentralized by design, built on the principle that mathematical work should be judged by the strength, rigor, and verifiability of the proof itself, not by the reputation of whoever submits it. Awards are paid in the recipient's choice of USDT on TRON (TRC-20) or USDC on Ethereum (ERC-20). Details on the prize, its problem catalog, and individual contributions are public through the program's GitHub repository.
Sun set up the prize in his own name as a long-term commitment to return wealth created through mathematics and technology back to mathematics itself. The Erdős catalog, maintained by Thomas Bloom, a mathematician and Royal Society University Research Fellow at the University of Manchester, holds more than 1,200 problems.
Who won, and for what
Van Doorn is an independent number theorist who started researching mathematics as an undergraduate in 2010 and kept publishing after leaving academia following his master's degree. He produced computer-checkable proofs in Lean for Erdős Problems #369, #457, and #469 — questions about consecutive integers with restricted prime factors, short runs of consecutive integers that collectively contain every prime in a given range, and whether reciprocals of a special class of numbers expressible as sums of their divisors add to a finite total.
Tang is a mathematics Ph.D. student at the University of Science and Technology of China. His research spans number theory, combinatorics, and AI-assisted mathematical discovery. He separately resolved Erdős Problem #1044, establishing a sharp lower limit for boundary lengths of regions defined by polynomials, and contributed alongside Li to a wider team's solution of Erdős Problem #1196, bounding weighted sums over sets of integers in which no member divides another.
Li is a mathematics researcher at Southeast University in Nanjing. He worked with Van Doorn and Tang on Erdős Problem #650, which asked exactly how many integers can always be matched to distinct multiples within a specified interval.
Where AI fit into the work
Problem #650 offers a concrete example of humans and AI splitting the labor. ChatGPT helped develop the proof strategy. Aristotle, an AI system built for mathematical reasoning, repaired a gap that appeared during Lean formalization. The researchers then simplified the argument and wrote the final proofs and exposition themselves.
Tang said the process changed how he thinks about collaboration. "This experience taught me how public feedback can sharpen a research question, and how AI-assisted discovery can combine mathematical judgment, collaboration and rigorous verification."
That's the quiet point of the prize: formal verification tools like Lean don't just check answers, they expose the holes a human proof can paper over. Having an AI system close one of those holes, and having a person write the exposition, is about as clean a division of labor as this kind of work gets.
More awards are expected as the program's catalog keeps moving. Erdős's problems are old, but the requirement that solutions be independently checkable is new — and it changes what counts as done. The repository remains the place to watch for the next confirmed winners.
