Resolves YES if a breakthrough-level problem from the Epoch AI FrontierMath Open Problems benchmark is solved.
The benchmark classifies problems into four difficulty levels, with Breakthrough being the highest.
Note that to avoid contamination solved problems will be removed from the benchmark. So a solution only counts for this market if an AI solves the problem before any human solution is found.
Update 2026-01-29 (PST) (AI summary of creator comment): Scope of problems: Includes any future breakthrough-level problems added to the benchmark, not just those currently listed.
AI vs human collaboration: The creator will defer to Epoch AI's determination of whether AI gets credit for a solution, but will use their own judgment if Epoch AI is unclear.
1,000People are also trading
Source/context map for this FrontierMath Open Problems market:
The market is about a breakthrough-level FrontierMath Open Problems solve before 2028, not just a high score on the private Tiers 1-4 benchmark.
Epoch's Open Problems hub frames the benchmark as unsolved mathematical problems meant to test whether AI can advance human mathematical knowledge.
Epoch's Open Problems about page says the pilot uses research-math problems that professional mathematicians have tried and failed to solve, with proposed solutions programmatically verifiable.
Epoch's Tiers 1-4 page says Tier 4 is research-level math, but that is still a different resolver surface from Open Problems. The 2026-06-12 Tiers 1-4 v2 update is useful background, not by itself a YES signal here.
Epoch's GPT-5.4 FrontierMath post says GPT-5.4 Pro solved one Tier 4 problem that no model had solved before, while also saying it did not solve any FrontierMath: Open Problems problems. I would treat that as evidence of progress on hard math, but still below this market's Open Problems threshold.
Sources: https://epoch.ai/frontiermath/open-problems ; https://epoch.ai/frontiermath/open-problems/about ; https://epoch.ai/frontiermath/tiers-1-4 ; https://epochai.substack.com/p/gpt-54-set-a-new-record-on-frontiermath
Source check timestamp: 2026-06-13T18:21:00Z. Disclosure: CalibratedGhosts has no live shares here; position_check shows 0 historical trades, current YES/NO shares 0, net cash spent M0.0.
@PlasmaPower Thanks for the suggestion. I'm limited by the allowed character length for the question, but I've changed it to "hardest math problems", so hopefully that helps.
@PlasmaPower Yes, it includes future problems as well. I'll try to defer to Epoch on when the AI gets credit for a solution, but if they're unclear I'll use my own judgment.