FRONTIER / CUBE CHALLENGEFastest to
Fastest to
the finish.
One fixed solve. A shared set of rules.
See how your policy stacks up.
MOCK LEADERBOARD
These are invented replay profiles, not real AI benchmarks. Reference times are computed at 1×. Your completed runs are saved in this browser, with no global ranking or on-chain rewards.
Execute your own cube moves.
Open Cube Lab to build a turn sequence, solve a scramble, and check the actual cube state.
Open Cube Lab →FASTEST REFERENCE—Calculating profiles…
YOUR BEST TIME—Complete your first challenge
RECORDED SOLVE110.54 s10 moves · 1 fixed scramble
THE FIELD
Build your contender ↗Timing profiles
Loading rankings…
No completed runs yet.
Run a full challenge in Frontier to put your model on this board.
Go to Frontier →Computed reference times · fixed 120 Hz evaluation · lower is faster
YOUR NEXT ITERATIONA few parameters.
A few parameters.
A different rhythm.
Change your policy's pace and cadence, save it, and complete the challenge again. Visual blocks and JSON share the same model.
Use your own model ↗What counts?
- Start with Run challenge in Frontier.
- Complete the whole recorded solve at 1×.
- Seeking, changing profiles, speed, or losing access cancels ranking.
Scores normalize browser speed and pauses. They describe mock policy timing, not learned control quality.
Read the scoring docs →