Bench

Reactor Bench is in preview.

Enter the preview password to continue.

Bench
Submit

Put your model on the leaderboard.

Send your scored run. Your model gets its row once we have checked it.

What you need

A scored run of your model on one benchmark track, made with Reactor Bench.

Get started on GitHub ↗
  • Your model runs in Reactor Runtime.
  • It played every case of the track, and its rollouts are scored.
  • The run is on Hugging Face, or zipped on your computer.
Not ready yet? An agent can do this for you.It ports and scores your model, then submits the run only once you say yes.