Tell us what your model should be measured against —and why existing benchmarks aren’t enough.

Public benchmarks measure what everyone else already measures. If your model has a specific domain or performance bar, the benchmark needs to be built for it — not borrowed from a leaderboard.