OpenAI GPT score on FrontierMath Benchmark by June 30?

Well... Typical polymarket here, when rules don't cover the situation at all. Basically all models got +42% for free due to errors in tasks. It doesn't reflect the nature of the market, when models were supposed to get improved... not the benchmark. Well there is now v1 published, which will resolve to No everywhere and v2, which will resolve to Yes otherwise. Rules don't state anything about 2 versions of the benchmark. I think it would be fair to resolve it 50-50 here in these circumstances. But I won't insist.

Should just resolve on v1 imo

When 42% of tasks are changed it is literally a new benchmark, not a little upgrade. Can appeal to the part of the rules that state "Studies which are not included in the leaderboard will not be considered". But I guess the opposite side will also find arguments. Rules are really not complete for this situation. I think 50-50 is the most fair option in this situation.

no way, the guy who bought hundreds of dollars of shares doesnt want to lose! FYI, this error was known for months and they said they would fix it soon.

Too late have I found out that the benchmark is rigged in favor of OpenAI.

guys my dog has GDV and things are looking bad.

My hedgehog has syphilis and things are looking even worse.

my dog got GDV and needs immediate care, please anything helps, i am so desperate.