Puzzle Solving Model Ranking

See which AI models perform best on Puzzle Solving, which ones stay reliable, and where the biggest gaps appear.

Models Shown

Average Puzzle Solving Score

0.7

Best Model

Failure Reasons

With failure reason Wrong answer344 With failure reason Did not follow instructions160 With failure reason API error35 With failure reason Extra formatting16 With failure reason No answer14 With failure reason Timed out11

442/442

Rank	Model	Company	Puzzle Solving Score	Score	Total Cost	Tests Correct	Response Time (avg)
#434	D1 default	Liquid	0.2	0.2	$0.003	0/1	570ms
Total Tests 1 Wrong Tests 1 Total Cost $0.003 Response Time (avg) 570ms
#438	Microsoft-Decision-1 default	Microsoft	0.2	0.1	$0.002	0/1	669ms
Total Tests 1 Wrong Tests 1 Total Cost $0.002 Response Time (avg) 669ms
#440	Tev1 4B Experimental default	Togethercomputer	0.2	1.0	$0.002	0/1	432ms
Total Tests 1 Wrong Tests 1 Total Cost $0.002 Response Time (avg) 432ms
#441	Nimble v3 9B default	Bespoke Labs	0.2	0.8	~$0.001	0/1	79ms
Total Tests 1 Wrong Tests 1 Total Cost ~$0.001 Response Time (avg) 79ms
#433	Step 3.5 Flash none	Stepfun	0.0	0.2	$0.020	0/0	0ms
Total Tests 0 Wrong Tests 0 Total Cost $0.020 Response Time (avg) 0ms
#435	Gemini 3 Flash Preview high	Google	0.0	0.2	$0.105	0/0	0ms
Total Tests 0 Wrong Tests 0 Total Cost $0.105 Response Time (avg) 0ms
#442	Claude Opus 5.5 none	Anthropic	0.0	0.2	$0.015	0/0	0ms
Total Tests 0 Wrong Tests 0 Total Cost $0.015 Response Time (avg) 0ms

Puzzle Solving Ranking