Skip to content
decisionbench
OverallSudokuRulebookConsistencyOdds

Independent benchmark. Not affiliated with or endorsed by OpenRouter or any model provider. Company and model names are trademarks of their respective owners.

How ranking worksPrivacyTermsContact

© 2026 editid0

← All races
← hard-01hard-03 →

Hard #02 sudoku race

27 clues · seed 3002 · 19 models · suite v1

19 AI decision models raced to solve this hard sudoku (27 clues), one digit at a time. Jev 1.14 by TypeSafe came first with 9 mistakes.

0.0s / 64.1s0/19 done

Results

#ModelMistakesCallsCostSolve timeStatus
1Jev 1.14TypeSafe924$0.0053727.2sSolved
2ClefCloudflare925$0.00952520sSolved
3Drex v1.5Nace.AIolder1125$0.00295224sSolved
4d1 UltraLiquid1227$0.0031799.6sSolved
5Clef OmniCloudflare1227$0.026711sSolved
6Span-02Respan1225$0.00414256sSolved
7d1Liquid1327$0.0015079.1sSolved
8Mercury DecideInception1327$0.00320419sSolved
9Decider Flash v1.2Perplexity1429$0.00529043sSolved
10Jev 1.13TypeSafeolder1529$0.0066008.5sSolved
11GPT-6 Luna DecisionsOpenAI1531$0.019212sSolved
12Kev 4BJared Palmer1530$0.00377419sSolved
13Tev1 4B ExperimentalTogether1629$0.024850sSolved
14Clef FlashCloudflare1833$0.00600119sSolved
15Drex v2.0Nace.AI1831$0.00698431sSolved
16Classify-1ScaleDown1832$0.034540sSolved
17Decider v1.2Perplexity1931$0.00240032sSolved
18Decision-1Microsoft2134$0.00459620sSolved
19Solar Decide FlashUpstage2539$0.058364sSolved