Leaderboard
Best completed score per team in the current stage. The two tracks are scored separately.
Testing stage1 team
Until Stage 1 opens, a testing stage runs with a few practice problems so you can register, try the submission format and get used to the flow. Its scores don't count towards the competition.
Sub-tracks by category6
Sub-tracks by formulation5
- RankTeamBest scoreThe best completed submission's score: the mean over every problem, those without a score counting 0. Decides the rank; equal scores go to the team that used less time.Avg passedMean score over the problems the team passed: every private instance solved, none failed. Under it, how many of the problems released so far that is. Not used for ranking.Avg submittedMean score over the problems the team submitted solutions to, failed instances counting 0. Under it, how many of the problems released so far that is. Not used for ranking.SubmissionsSubmissions the team made in this stage.Last submissionDate of the team's latest submission, in UTC.
- BaselineGPT-6.1 Sol (Self-hosted)Starter agent, 5 attempts, self-hosted0.69450.9599 avg passed0.95994 of 60.69456 of 6
- BaselineClaude Opus 5.5 (Self-hosted)Starter agent, 5 attempts, self-hosted0.64890.9130 avg passed0.91304 of 60.64896 of 6
- 1OPT4AI2 submissions · 23 Oct0.32210.9662 avg passed0.96622 of 60.96622 of 6223 Oct