Submission form

Please send up problems using this form, and our mathematicians will evaluate how difficult they are relative to what we’re looking for, and we’ll compensate you $300-$1000 based on those assessments. We will contact you via the email you provide.

Note that we reserve the right to decline purchasing a question even if it fits the criteria, due to high submission volume or other processing constraints such as limitations in contracting in certain countries.

We will also ask you to sign a light NDA agreement before we purchase your question, to ensure the questions remain confidential.

Problem requirements

• The problems should be exceptionally challenging, with a strong emphasis on research-like questions. We’re particularly seeking problems that mirror those encountered in actual mathematical research, requiring deep specialized knowledge and potentially taking experts hours to solve.
• We recommend you to use OpenAI’s latest o1-mini model on chatgpt.com (or 4o-mini if you are a free user) to test the difficulty level of your problems before submission. Any problem that o1-mini gets right in at least 1 out of 3 attempts is very likely to be rejected. If you do so, please enter only the problem statement, not the answer. Furthermore, there is an option to prevent the model from training on your text: click the ? in the lower right corner, then settings, then deselect “Improve the model for everyone.”
• The questions should have a clear answer that can be automatically verified. Ideally this answer is an integer or a rational number, but we can accommodate more general type signatures of answers so long as the model can write a Python script that computes the answer and that can be checked for correctness using test cases. Consequently, e.g. questions of the type “Prove that…” should be avoided.
• Guessing the right answer to the question with a > 1% chance should be as difficult as solving the problem. As a result, true/false questions or questions where most of the difficulty is in proving that a conjectured answer is correct should be avoided.

For more info please see these guidelines.