@DottedCalculator mind explaining your big NO bet? Did it somehow emerge that the problems were harder than you thought?

@AhronMaline I don’t want my entire net worth locked up in the market, so I’m selling still at a large profit (I bought a lot at 60)

@AhronMaline note that he's still the biggest YES holder on /johnNZOy/gemini-perfect-score-on-imo-2026 at 80%

should be resolved

GitHub - AxiomMath/IMO2026

GitHub - AxiomMath/IMO2026

Contribute to AxiomMath/IMO2026 development by creating an account on GitHub.

@AnirudhBharadwajRhu5 No, this doesn't qualify! Look at the time reported for each question; it's well beyond the IMO time limits (270 minutes for each set of 3 problems)

im pretty sure it has already been confirmed that this happened

source?

5.6 Pro basically did 42/42 in my tests

@mathvc omegalol

@mathvc https://github.com/SignalPilot-Labs/AutoFyn/blob/production/results/imo-2026/pdfs/IMO_performance_by_GPT_5_6_sol.pdf

I heard GPT 5.6 was able to solve P3

Very easy test. CG/A/N

combo flavored N/G/C. Easier than 2025 day 2. G is annoying to directly bash but it’s easy to reduce to something bashable with computer

Day 1 is today!

@DottedCalculator when are the problems typically publicly known?

@Bayesian 1:00+the time it takes to get the contestants out of the room+the time it takes for the contestants to post the problems online = around 1:30 pm (in 7.5 hours). I can probably post the subject distribution here but I probably won’t be the one to post the problems online.

the imo and the world cup semi finals are on the same dates

@121 in fact the IMO provides World Cup semifinals replay screenings as an official activity after each day

priced reasonably at 88% IMO

@jim this rly feels high holy moly

@Bayesian it's cool to see manifold more optimistic than jim on this. i rmeember last year i was an outlier thinking LLMs would get gold at all

@jim tbf they got a bit lucky last year

@jim I still think it’s too low but I’m out of mana now

@Bayesian I also think they got lucky last year, but 1 year of progress is a lot, and OpenAI clearly cares

priced way too low IMO.