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.

GB2-02: Chess - Reasoning Variants
Rank Model Reasoning Score Programs Raw rating Combined record (W / D / L)
1Claude Fable 5.2XHigh99.61 working of 1 attempt2546.1559 / 3 / 0
2Gemini 4 HydrogenXHigh96.61 working of 1 attempt2157.7521 / 37 / 4
3Claude Sonnet 5.5XHigh96.01 working of 1 attempt2123.3518 / 35 / 9
4Gemini 4 CarbonHigh96.01 working of 1 attempt2123.3525 / 21 / 16
5Claude Opus 5.5XHigh95.61 working of 1 attempt2104.0517 / 32 / 13
6Claude Fable 5.1XHigh95.11 working of 1 attempt2082.5516 / 28 / 18
7GPT-6 AstraXHigh94.81 working of 1 attempt2065.8510 / 35 / 17
8Claude Opus 5.5Medium94.31 working of 1 attempt2047.1511 / 27 / 24
9GPT-6.1 SolXHigh94.11 working of 1 attempt2038.1499 / 48 / 15
10Gemini 4 ArgonHigh93.91 working of 1 attempt2032.4491 / 62 / 9
11Claude Fable 5.2Medium93.81 working of 1 attempt2029.5506 / 31 / 25
12Gemini 4 CarbonMedium91.11 working of 1 attempt1946.7468 / 73 / 21
13Pareto 26.10 PreviewDefault90.61 working of 1 attempt1934.3477 / 49 / 36
14Claude Fable 5.1Medium90.31 working of 1 attempt1926.3477 / 45 / 40
15GPT-6.1 Terra ProXHigh88.71 working of 1 attempt1891.1461 / 58 / 43
16Grok 4.7XHigh87.71 working of 1 attempt1870.8453 / 62 / 47
17Gemini 4 CarbonLow86.71 working of 1 attempt1850.2440 / 75 / 47
18Claude Opus 5XHigh82.62 working of 2 attempts1789.6 program range 1727–1852794 / 226 / 104
19GPT-6 SolXHigh82.01 working of 1 attempt1774.1367 / 165 / 30
20GLM-5.3 FlashX LegalMax82.01 working of 1 attempt1774.1407 / 85 / 70
21Kimi K3.1 ProLow81.81 working of 1 attempt1771.6383 / 131 / 48
22Grok 4.6XHigh81.71 working of 1 attempt1770.4402 / 92 / 68
23Claude Fable 5.1Low81.01 working of 1 attempt1760.9371 / 146 / 45
24GPT-6 AstraMedium80.81 working of 1 attempt1757.4351 / 183 / 28
25DeepSeek V4.1 Flash 1017Max79.71 working of 1 attempt1742.7377 / 118 / 67

Individual player programs

Each row is one accepted player program for GB2-02: Chess; 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.

280 player programs.

ModelReasoningStatusScoreRatingW / D / LMatchesOpponentsNotesGenerated
Claude Fable 5.2E-56812c272561d614XHighRated Interval validated99.62546559 / 3 / 0562 / 562281—2026-11-09First try
Gemini 4 HydrogenE-e82ed6b9c68ae82bXHighRated Interval validated96.62158521 / 37 / 4562 / 562281—2026-11-12First try
Claude Sonnet 5.5E-25c4c1579034e616XHighRated Interval validated96.02123518 / 35 / 9562 / 562281—2026-09-30First try
Gemini 4 CarbonE-3a5ab7a8b8473deaHighRated Interval validated96.02123525 / 21 / 16562 / 562281—2026-11-12Repaired
Claude Opus 5.5E-8da2a605b97075faXHighRated Interval validated95.62104517 / 32 / 13562 / 562281—2026-09-22First try
Claude Fable 5.1E-f86f2f46e2c71b4aXHighRated Interval validated95.12082516 / 28 / 18562 / 562281—2026-09-08First try
GPT-6 AstraE-c91ee6859ab8877eXHighRated Interval validated94.82066510 / 35 / 17562 / 562281—2026-09-05First try
Claude Opus 5.5E-4692bc19babebbb1MediumRated Interval validated94.32047511 / 27 / 24562 / 562281—2026-09-22Repaired
GPT-6.1 SolE-755187947384baf1XHighRated Interval validated94.12038499 / 48 / 15562 / 562281—2026-10-01First try
Gemini 4 ArgonE-35eb3b86b76e7c9fHighRated Interval validated93.92032491 / 62 / 9562 / 562281—2026-10-21Repaired
Claude Fable 5.2E-692854324abe471cMediumRated Interval validated93.82030506 / 31 / 25562 / 562281—2026-11-09First try
Gemini 4 CarbonE-207a6d6cbd1d627aMediumRated Interval validated91.11947468 / 73 / 21562 / 562281—2026-11-12First try
Pareto 26.10 PreviewE-e13381315ea87cb7DefaultRated Interval validated90.61934477 / 49 / 36562 / 562281—2026-10-09First try
Claude Fable 5.1E-f4d4ba39a6f2270eMediumRated Interval validated90.31926477 / 45 / 40562 / 562281—2026-09-03First try
GPT-6.1 Terra ProE-26c8994965e09404XHighRated Interval validated88.71891461 / 58 / 43562 / 5632811 fault2026-11-01First try
Grok 4.7E-ba03a6f2135ea911XHighRated Interval validated87.71871453 / 62 / 47562 / 562281—2026-09-30First try
Claude Opus 5E-fcdd98d74e80134aXHighRated Interval validated86.71852455 / 46 / 61562 / 562281—2026-07-25First try
Gemini 4 CarbonE-6a27e144b3514939LowRated Interval validated86.71850440 / 75 / 47562 / 562281—2026-11-12First try
GLM-5.3 FlashX LegalE-b4431d08cc503c1eMaxRated Interval validated82.01774407 / 85 / 70562 / 562281—2026-11-03First try
GPT-6 SolE-6b0e5cea9d2e9834XHighRated Interval validated82.01774367 / 165 / 30562 / 562281—2026-09-22First try
Kimi K3.1 ProE-313ef2e01b0b5aa0LowRated Interval validated81.81772383 / 131 / 48562 / 562281—2026-10-24First try
Grok 4.6E-892d7d5eef11e96cXHighRated Interval validated81.71770402 / 92 / 68562 / 562281—2026-09-03First try
Claude Fable 5.1E-85e618bbeae3eac9LowRated Interval validated81.01761371 / 146 / 45562 / 562281—2026-09-03First try
GPT-6 AstraE-db1b6aef6e22ec35MediumRated Interval validated80.81757351 / 183 / 28562 / 562281—2026-09-05First try
DeepSeek V4.1 Flash 1017E-5e795839a2bce09cMaxRated Interval validated79.71743377 / 118 / 67562 / 562281—2026-10-20First try