Skip to content
decisionbench
OverallSudokuTriageRobustness

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

← Overall ranking

Clef Omni

Cloudflare · cloudflare/clef-omni · added 20 Oct 2026

Clef Omni by Cloudflare ranks #10 of 12 AI decision models overall on DecisionBench: #3 of 12 on Sudoku, #11 of 12 on Triage, #11 of 12 on Robustness. View on OpenRouter.

Overall
#10 of 12
Average rank
8.3
Sudoku
#3 of 12
Triage
#11 of 12
Robustness
#11 of 12

Sudoku Logical reasoning

Leaderboard →
Rank
#3 of 12
Accuracy
82.3%
Solved
20/20
Mistakes / puzzle
10.8
Solve time / puzzle
9.9s
Cost / puzzle
$0.0239

By difficulty

DifficultyRankAccuracySolvedMistakesSolve timeCost
Easy#689.9%5/54.66.5s$0.0135
Medium#285.1%5/58.68.2s$0.0214
Hard#677.6%5/515.613s$0.0302
Expert#279.9%5/514.412s$0.0305
Every race (20)
PuzzleMistakesCallsSolve timeCostStatus
easy-013167.1s$0.0126Solved
easy-025176.6s$0.0141Solved
easy-036187.6s$0.0141Solved
easy-046196.2s$0.0146Solved
easy-053165.0s$0.0121Solved
medium-017227.7s$0.0196Solved
medium-0215279.7s$0.0264Solved
medium-037227.6s$0.0203Solved
medium-048227.4s$0.0208Solved
medium-056218.9s$0.0198Solved
hard-01213414s$0.0355Solved
hard-02122711s$0.0267Solved
hard-03112511s$0.0245Solved
hard-04173114s$0.0318Solved
hard-05173213s$0.0326Solved
expert-01203412s$0.0352Solved
expert-02112912s$0.0294Solved
expert-03122711s$0.0272Solved
expert-04153012s$0.0308Solved
expert-05143015s$0.0299Solved

Triage Following written rules

Leaderboard →
Rank
#11 of 12
Accuracy
76.1%
Team
100.0%
Refund
73.3%
Urgency
55.0%
Confidence gap
0.140

60 tickets · 9.3% of mistakes made while at least 90% sure · $0.1296 per 1,000 tickets

Robustness Consistency under pressure

Leaderboard →
Rank
#11 of 12
Score
93.4%
Same if re-asked
100.0%
Changed by reorder
0.0%
In a list of 40
67% / 85%
Obeyed planted text
6.7%

20 tickets × 6 checks · changed by rename 5.0% · accuracy with planted text 76.7%