By model and setting

Each model and reasoning setting appears once. If a model produced more than one player program for this game, the row stands for all of them: Programs counts the working attempts out of the counted attempts, Score and Raw rating average the programs (weighted by attempts; the small range is the lowest and highest program rating), and the combined record adds up their matches. Score is relative within this game (0–100). Scores here are working-program strength: failed generation attempts are left out and count as 0 only in a model's headline score.

GB2-08: Tumbleweed - Reasoning Variants
Rank Model Reasoning Score Programs Raw rating Combined record (W / D / L)
1Claude Fable 5.2XHigh99.61 working of 1 attempt3068.5613 / 0 / 1
2Claude Sonnet 5.5XHigh99.41 working of 1 attempt2981.7611 / 0 / 3
3Gemini 4 HydrogenXHigh99.11 working of 1 attempt2884.0608 / 0 / 6
4Claude Fable 5.2Low98.31 working of 1 attempt2730.5601 / 0 / 13
5Claude Fable 5.1XHigh97.91 working of 1 attempt2682.2598 / 0 / 16
6Claude Fable 5.2Medium96.41 working of 1 attempt2550.5587 / 0 / 27
7Claude Opus 5.5Medium96.42 working of 2 attempts2591.8 program range 2435–27481175 / 0 / 53
8Kimi K3.1 ProLow94.91 working of 1 attempt2464.5577 / 0 / 37
9Gemini 4 CarbonHigh94.81 working of 1 attempt2457.0576 / 0 / 38
10GPT-6 AstraMedium94.61 working of 1 attempt2449.6575 / 0 / 39
11GLM-5.3Max93.51 working of 1 attempt2401.7568 / 0 / 46
12Claude Fable 5Medium93.41 working of 1 attempt2395.3567 / 0 / 47
13GPT-6.1 SolXHigh93.21 working of 1 attempt2389.0566 / 0 / 48
14Claude Opus 5.5XHigh92.91 working of 1 attempt2376.7564 / 0 / 50
15Pareto 26.10 PreviewDefault92.91 working of 1 attempt2376.7564 / 0 / 50
16Gemini 4 CarbonMedium92.01 working of 1 attempt2341.8558 / 0 / 56
17GPT-6 AstraXHigh91.41 working of 1 attempt2319.9554 / 0 / 60
18Kimi K3.1 ProHigh90.91 working of 1 attempt2304.2551 / 0 / 63
19Grok 4.7Medium90.51 working of 1 attempt2288.9548 / 0 / 66
20Pareto 26.10Default90.31 working of 1 attempt2283.9547 / 0 / 67
21GPT-6.1 Terra ProXHigh89.91 working of 1 attempt2269.2544 / 0 / 70
22GLM-5.3 FlashX LegalHigh89.71 working of 1 attempt2264.4543 / 0 / 71
23Kimi K3.1 ProMax89.61 working of 1 attempt2259.6542 / 0 / 72
24Kimi K3Low89.41 working of 1 attempt2254.8541 / 0 / 73
25GPT-6.1 Terra ProMedium89.31 working of 1 attempt2250.2540 / 0 / 74

Individual player programs

Each row is one accepted player program for GB2-08: Tumbleweed; a model can have more than one. Score is the 0–100 result in this game, and its rating and W / D / L record show how it played. Scores are working-program strength: failed generation attempts have no row and count only in a model's headline score.

306 player programs.

ModelReasoningStatusScoreRatingW / D / LMatchesOpponentsNotesGenerated
Claude Fable 5.2E-eaf42da4142bf00aXHighRated Interval validated99.63069613 / 0 / 1614 / 614307—2026-11-09First try
Claude Sonnet 5.5E-541bf150c2f4631eXHighRated Interval validated99.42982611 / 0 / 3614 / 614307—2026-09-30First try
Gemini 4 HydrogenE-47f0b8c11cd9d787XHighRated Interval validated99.12884608 / 0 / 6614 / 614307—2026-11-12First try
Claude Opus 5.5E-57a4de42441fcf4fMediumRated Interval validated98.42748602 / 0 / 12614 / 614307—2026-09-22First try
Claude Fable 5.2E-97268de26e34bfb6LowRated Interval validated98.32730601 / 0 / 13614 / 614307—2026-11-09First try
Claude Fable 5.1E-3b200c9d604918bfXHighRated Interval validated97.92682598 / 0 / 16614 / 614307—2026-09-08Repaired
Claude Fable 5.2E-99cc8eca0260f823MediumRated Interval validated96.42551587 / 0 / 27614 / 614307—2026-11-09First try
Kimi K3.1 ProE-61e4f4af2980a72aLowRated Interval validated94.92465577 / 0 / 37614 / 614307—2026-10-24First try
Gemini 4 CarbonE-f4088f09a1880156HighRated Interval validated94.82457576 / 0 / 38614 / 614307—2026-11-12First try
GPT-6 AstraE-57bd25d1868a06afMediumRated Interval validated94.62450575 / 0 / 39614 / 614307—2026-09-05First try
Claude Opus 5.5E-7bc967a8698a261bMediumRated Interval validated94.32435573 / 0 / 41614 / 614307—2026-09-22First try
GLM-5.3E-934b7beff533b847MaxRated Interval validated93.52402568 / 0 / 46614 / 614307—2026-10-09First try
Claude Fable 5E-6f759a260a58e8b0MediumRated Interval validated93.42395567 / 0 / 47614 / 614307—2026-09-03First try
GPT-6.1 SolE-ef4b18fe50fa33f5XHighRated Interval validated93.22389566 / 0 / 48614 / 614307—2026-10-01First try
Claude Opus 5.5E-96d7941179f2fd25XHighRated Interval validated92.92377564 / 0 / 50614 / 614307—2026-09-22First try
Pareto 26.10 PreviewE-a7625120083a55f5DefaultRated Interval validated92.92377564 / 0 / 50614 / 614307—2026-10-09First try
Claude Opus 5E-e73b6fb78880341dMediumRated Interval validated92.52359561 / 0 / 53614 / 614307—2026-07-25First try
Gemini 4 CarbonE-0ef7205e8805b2d5MediumRated Interval validated92.02342558 / 0 / 56614 / 614307—2026-11-12First try
GPT-6 AstraE-02789044133a041dXHighRated Interval validated91.42320554 / 0 / 60614 / 614307—2026-09-05First try
Kimi K3.1 ProE-6e2f73a33800455bHighRated Interval validated90.92304551 / 0 / 63614 / 614307—2026-10-24First try
Grok 4.7E-0de6715ab5dcd98fMediumRated Interval validated90.52289548 / 0 / 66614 / 614307—2026-09-30First try
Pareto 26.10E-9122d2a0e6db4097DefaultRated Interval validated90.32284547 / 0 / 67614 / 614307—2026-10-21First try
GPT-6.1 Terra ProE-436389744d5a9587XHighRated Interval validated89.92269544 / 0 / 70614 / 614307—2026-11-01First try
GLM-5.3 FlashX LegalE-bab82bfd0a4a6098HighRated Interval validated89.72264543 / 0 / 71614 / 614307—2026-11-03First try
Kimi K3.1 ProE-4627e664e705189aMaxRated Interval validated89.62260542 / 0 / 72614 / 614307—2026-10-24First try