Claude Sonnet 5.5 Medium
Answered each puzzle once; the other models answered each 3 times. Fewer samples make this score noisier: a rerun could land several points higher or lower, so read its rank as approximate.
27.4Overall
By game
GameQualityLegalityCostnRank
Xiangqi33.5100.0$0.0131 of 4
xiangqi-01e4e3100.0100.0$0.01
xiangqi-02b4d50.0100.0$0.01
xiangqi-03h4h60.6100.0$0.01
Chess0.066.7$0.0132 of 4
chess-01Bxf6 (illegal)0.00.0$0.01
chess-02e4f30.0100.0$0.01
chess-03g8f80.0100.0$0.01
Go38.7100.0$0.0132 of 4
go-01E414.2100.0$0.01
go-02S6100.0100.0$0.01
go-03E111.9100.0$0.01
Backgammon26.3100.0$0.0134 of 4
backgammon-019/3 5/40.0100.0$0.01
backgammon-0213/8 2/145.4100.0$0.01
backgammon-036/1 5/133.3100.0$0.01
Shogi66.766.7$0.0132 of 4
shogi-014d4c+100.0100.0$0.01
shogi-02P*3c (illegal)0.00.0$0.01
shogi-035f5e100.0100.0$0.01
Scrabble13.133.3$0.0133 of 4
scrabble-01J8 JEST39.2100.0$0.01
scrabble-028D ZAiRES (illegal)0.00.0$0.01
scrabble-038I ED (illegal)0.00.0$0.02
Kalah33.3100.0$0.0133 of 4
kalah-016100.0100.0$0.01
kalah-0250.0100.0$0.01
kalah-0350.0100.0$0.01
Gomoku14.0100.0$0.0134 of 4
gomoku-01g937.5100.0$0.01
gomoku-02j84.4100.0$0.01
gomoku-03h80.0100.0$0.01
Othello1.2100.0$0.0134 of 4
othello-01h70.0100.0$0.01
othello-02g80.0100.0$0.01
othello-03h33.6100.0$0.01
Connect Four47.1100.0$0.0132 of 4
connect4-01541.3100.0$0.01
connect4-025100.0100.0$0.01
connect4-0370.0100.0$0.01
Quality is the mean move score × 100. n is answers: puzzles × samples. Open a game for each puzzle and the move played. Protocol boardbench-0.2.