Leaderboard
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.
| Rank | Model | Reasoning | Score | Programs | Raw rating | Combined record (W / D / L) |
|---|---|---|---|---|---|---|
| 1 |  GPT-6.1 Sol | XHigh | 95.1 | 1 working of 1 attempt | 2073.8 | 455 / 36 / 7 |
| 2 |  Claude Opus 5.5 | XHigh | 94.8 | 1 working of 1 attempt | 2062.7 | 450 / 43 / 5 |
| 3 | GPT-6 Astra | XHigh | 92.6 | 1 working of 1 attempt | 1988.4 | 443 / 33 / 22 |
| 4 |  Grok 4.7 | Medium | 92.3 | 1 working of 1 attempt | 1980.5 | 436 / 44 / 18 |
| 5 | GPT-6 Astra | Medium | 91.4 | 1 working of 1 attempt | 1955.6 | 431 / 44 / 23 |
| 6 | Claude Opus 5.5 | Medium | 88.5 | 1 working of 1 attempt | 1890.4 | 417 / 41 / 40 |
| 7 | GPT-6 Sol | XHigh | 88.3 | 1 working of 1 attempt | 1884.8 | 409 / 54 / 35 |
| 8 | Claude Fable 5.1 | XHigh | 88.0 | 2 working of 2 attempts | 1883.4 program range 1834–1933 | 824 / 90 / 82 |
| 9 | GPT-6.1 Sol | Medium | 87.2 | 1 working of 1 attempt | 1863.5 | 411 / 38 / 49 |
| 10 |  Pareto 26.10 Preview | Provider | 86.4 | 1 working of 1 attempt | 1850.0 | 404 / 44 / 50 |
| 11 | Pareto 26.10 | Provider | 85.7 | 1 working of 1 attempt | 1837.1 | 391 / 62 / 45 |
| 12 |  Qwen3.8 Max 0902 | XHigh | 84.7 | 1 working of 1 attempt | 1821.7 | 373 / 88 / 37 |
| 13 | Claude Fable 5 | XHigh | 84.2 | 1 working of 1 attempt | 1814.3 | 378 / 73 / 47 |
| 14 | GPT-5.6 Sol | XHigh | 83.9 | 1 working of 1 attempt | 1808.5 | 368 / 89 / 41 |
| 15 |  Kimi K3.1 Pro | Max | 83.8 | 1 working of 1 attempt | 1807.0 | 355 / 114 / 29 |
| 16 |  GLM-5.3 | High | 83.6 | 1 working of 1 attempt | 1803.9 | 362 / 96 / 39 |
| 17 | Claude Sonnet 5.5 | XHigh | 83.5 | 1 working of 1 attempt | 1802.7 | 365 / 91 / 42 |
| 18 | GPT-6 Astra | Low | 83.4 | 1 working of 1 attempt | 1801.3 | 368 / 84 / 46 |
| 19 | Kimi K3.1 Pro | High | 82.2 | 1 working of 1 attempt | 1783.4 | 353 / 101 / 44 |
| 20 |  Muse Wonder | XHigh | 81.9 | 1 working of 1 attempt | 1779.4 | 370 / 64 / 64 |
| 21 |  Gemini 3.1 Pro Preview | High | 80.4 | 1 working of 1 attempt | 1760.1 | 341 / 107 / 50 |
| 22 | Claude Opus 5 | XHigh | 80.4 | 2 working of 2 attempts | 1771.5 program range 1683–1860 | 707 / 165 / 124 |
| 23 | Claude Haiku 5.5 | XHigh | 80.0 | 1 working of 1 attempt | 1753.9 | 332 / 120 / 46 |
| 24 | Kimi K3.1 Pro | Low | 78.9 | 1 working of 1 attempt | 1740.5 | 323 / 127 / 48 |
| 25 | GLM-5.3 Flash | Max | 78.8 | 1 working of 1 attempt | 1739.3 | 317 / 138 / 43 |
Underlying results
Individual player programs
Each row is one accepted player program for GB2-04: Xiangqi; 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.
250 player programs.
| Model | Reasoning | Status | Score | Rating | W / D / L | Matches | Opponents | Notes | Generated |
|---|---|---|---|---|---|---|---|---|---|
GPT-6.1 SolE-0beb3abad220fddb | XHigh | Rated Interval validated | 95.1 | 2074 | 455 / 36 / 7 | 498 / 498 | 249 | — | 2026-10-01First try |
Claude Opus 5.5E-d512f5c88a2f7092 | XHigh | Rated Interval validated | 94.8 | 2063 | 450 / 43 / 5 | 498 / 498 | 249 | — | 2026-09-22Repaired |
GPT-6 AstraE-154ed79df2f33a60 | XHigh | Rated Interval validated | 92.6 | 1988 | 443 / 33 / 22 | 498 / 498 | 249 | — | 2026-09-05First try |
Grok 4.7E-3fd10aaafda5a40d | Medium | Rated Interval validated | 92.3 | 1980 | 436 / 44 / 18 | 498 / 498 | 249 | — | 2026-09-30First try |
GPT-6 AstraE-8863403f24a275fd | Medium | Rated Interval validated | 91.4 | 1956 | 431 / 44 / 23 | 498 / 498 | 249 | — | 2026-09-05First try |
Claude Fable 5.1E-6c73f19c641d4021 | XHigh | Rated Interval validated | 90.5 | 1933 | 421 / 54 / 23 | 498 / 498 | 249 | — | 2026-09-08First try |
Claude Opus 5.5E-d01a435c350a4a98 | Medium | Rated Interval validated | 88.5 | 1890 | 417 / 41 / 40 | 498 / 498 | 249 | — | 2026-09-22First try |
Ox AlphaE-454f79b70ee12fe0 | Max | Rated Interval validated | 88.4 | 1887 | 394 / 85 / 19 | 498 / 498 | 249 | — | 2026-08-22First try |
GPT-6 SolE-ca73d8cf5d48a8e6 | XHigh | Rated Interval validated | 88.3 | 1885 | 409 / 54 / 35 | 498 / 498 | 249 | — | 2026-09-22First try |
GPT-6.1 SolE-ccfb3a267deaea31 | Medium | Rated Interval validated | 87.2 | 1863 | 411 / 38 / 49 | 498 / 498 | 249 | — | 2026-10-01First try |
Claude Opus 5E-a02883202d334eec | XHigh | Rated Interval validated | 87.0 | 1860 | 387 / 84 / 27 | 498 / 498 | 249 | — | 2026-07-25First try |
Pareto 26.10 PreviewE-4d1e4650d78a4cb3 | Provider | Rated Interval validated | 86.4 | 1850 | 404 / 44 / 50 | 498 / 498 | 249 | — | 2026-10-09First try |
Pareto 26.10E-334062be20fe1540 | Provider | Rated Interval validated | 85.7 | 1837 | 391 / 62 / 45 | 498 / 498 | 249 | — | 2026-10-21First try |
Claude Fable 5.1E-7002caf3e34510f6 | XHigh | Rated Interval validated | 85.5 | 1834 | 403 / 36 / 59 | 498 / 498 | 249 | — | 2026-09-03First try |
Qwen3.8 Max 0902E-924ba680b1c3e7cd | XHigh | Rated Interval validated | 84.7 | 1822 | 373 / 88 / 37 | 498 / 498 | 249 | — | 2026-10-09First try |
Claude Fable 5E-e34f220d213bb5a8 | XHigh | Rated Interval validated | 84.2 | 1814 | 378 / 73 / 47 | 498 / 498 | 249 | — | 2026-09-03First try |
GPT-5.6 SolE-6b542c5d0dcf2279 | XHigh | Rated Interval validated | 83.9 | 1808 | 368 / 89 / 41 | 498 / 498 | 249 | — | 2026-08-09First try |
Kimi K3.1 ProE-8504f908530a492a | Max | Rated Interval validated | 83.8 | 1807 | 355 / 114 / 29 | 498 / 498 | 249 | — | 2026-10-24First try |
GLM-5.3E-c2bce9f934aadf7b | High | Rated | 83.6 | 1804 | 362 / 96 / 39 | 497 / 498 | 249 | 1 fault | 2026-09-03Repaired |
Claude Sonnet 5.5E-dcc5e8b06de17dda | XHigh | Rated Interval validated | 83.5 | 1803 | 365 / 91 / 42 | 498 / 498 | 249 | — | 2026-09-30Repaired |
GPT-6 AstraE-03f8ff0d1cf7b904 | Low | Rated Interval validated | 83.4 | 1801 | 368 / 84 / 46 | 498 / 498 | 249 | — | 2026-09-05First try |
Kimi K3.1 ProE-6c5309fb9d50203c | High | Rated Interval validated | 82.2 | 1783 | 353 / 101 / 44 | 498 / 498 | 249 | — | 2026-10-24First try |
Muse WonderE-ee60f0724f1a214b | XHigh | Rated Interval validated | 81.9 | 1779 | 370 / 64 / 64 | 498 / 498 | 249 | — | 2026-10-12First try |
Gemini 3.1 Pro PreviewE-f62bedab78302d4e | High | Rated Interval validated | 80.4 | 1760 | 341 / 107 / 50 | 498 / 498 | 249 | — | 2026-09-30First try |
Claude Haiku 5.5E-ca9b7f741dced799 | XHigh | Rated Interval validated | 80.0 | 1754 | 332 / 120 / 46 | 498 / 498 | 249 | — | 2026-10-09First try |