Coding benchmark v1

GPT-5.6-Luna (max)

Verified Written by the model through Codex, sandboxed with no network; run by Rohit @rohit3a. Graded 13 Oct 2026 04:22 UTC. All solvers

Score6.8100 = 60 moves a cube
Solved50of 50 hidden cubes
Moves1035.0per cube solved
Per cube6.5 slimit 60 s
Coding3h 21mof 4 h
Cost$3.04114.0M tokens

The 50 hidden cubes

50 solved

  1. 155010.9
  2. 241514.5
  3. 36778.9
  4. 412784.7
  5. 59616.2
  6. 610495.7
  7. 79816.1
  8. 89916.1
  9. 910305.8
  10. 108067.4
  11. 119656.2
  12. 126918.7
  13. 1313684.4
  14. 1411625.2
  15. 1516083.7
  16. 1613884.3
  17. 1713974.3
  18. 1812134.9
  19. 196489.3
  20. 2010865.5
  21. 2114054.3
  22. 2215693.8
  23. 236539.2
  24. 2442514.1
  25. 259146.6
  26. 2648412.4
  27. 2714804.1
  28. 286199.7
  29. 2911985.0
  30. 307827.7
  31. 319626.2
  32. 3238215.7
  33. 3310046.0
  34. 346619.1
  35. 3521872.7
  36. 3610125.9
  37. 376099.9
  38. 389316.4
  39. 399076.6
  40. 4017253.5
  41. 4116943.5
  42. 4211615.2
  43. 4311915.0
  44. 4412564.8
  45. 4510775.6
  46. 4613254.5
  47. 4715483.9
  48. 4810235.9
  49. 4959210.1
  50. 507128.4

Moves for each cube solved, and its score (100 × 60 ÷ moves). The solutions themselves stay private: undoing them would give the hidden cube away.

The solver

Frozen when the model stopped: 152 KB gzipped, sha256 b54cbba81b11d4d1…. Download the code