Bug Hunt Bench · Blind-graded · One prompt · Updated
Bug Hunt Bench
Choose runs
A model can appear more than once — different reasoning tiers, or a different harness. Runs tagged superseded are historical and stay out of the default view.
No runs selected. Pick some above, or hit Featured.
No runs selected. Pick some above, or hit Featured.
Cost is on a logarithmic axis: the spread across the board is roughly two hundredfold, and a linear axis would pile half the runs into the left margin. These dollar figures are not one kind of number. Each point carries its own tag — bill, list rate, floor, free — so horizontal distance between two differently tagged points is not a like-for-like comparison. The step line marks the cost–score frontier: runs that no other selected run beats on both axes at once. Every value here is also in the .
How it works
The method
Where it stops
What this board does not tell you
Read the columns right