No matching problems.
Try a shorter search or fewer filters.
We are collecting 500 important open problems in theoretical computer science. We want AI labs to work on them and use the collection as a benchmark of research capabilities, while advancing our field. Solutions will require Lean proofs so they can be checked automatically.
You can , comment on an entry, or vote on what should be included. No account is needed.
Try a shorter search or fewer filters.
Ideas from readers, open for discussion. New suggestions appear here immediately; we review them before adding them to the catalogue. Comments and references are welcome.
Loading shared contributions…
The catalogue brings together open questions across theoretical computer science, ordered within each category by votes, with catalogue priority breaking ties. Some formulations and their current open status still need review; each card preserves its sources and known limitations. Resolved and retired records are excluded from the reader.
Use Compact for a quick overview or Full for the complete statement and context. Anyone can suggest a problem or leave a public comment here without an account. Names are optional and unverified. Suggestions stay separate from the catalogue until reviewed. You can delete your own contributions from the browser you used to submit them.
Vote with 👍 or 👎; click your selected thumb again to remove your vote. One vote per problem is remembered in this browser. Scores are shared with everyone and determine the Top selections within each category.
Our primary goal is a benchmark of 500 problems. Top 100 is a priority subset of those same 500, with secondary editorial attention.