Combined x Wrong answer Ranking

See which AI models are most likely to hit Wrong answer on Combined, so you can spot weak points faster. Sort by: Total Cost ↑.

Models Shown

Total Failures

Most Affected Model

Failure Reasons

Invalid tool call91 Wrong answer68 No answer29 API error26 Timed out5 Did not follow instructions1 Extra formatting1

Categories

Domain specific412 Anti-AI Tricks293 Coding252 Puzzle Solving201 Trivia168 Combined68 Instructions following61 General Intelligence59 Data parsing and extraction41 Tool Calling3

63/63

Rank	Model	Company	Wrong answer Count	Category Score	Total Cost	Tests Correct	Response Time (avg)
#104	Gemini 3.1 Flash Lite Preview low	Google	1	3.0	$0.646	0/2	160.6s
Total Tests 2 Wrong Tests 2 Total Cost $0.646 Response Time (avg) 160.6s
#52	Kimi K2.7 Code medium	Moonshot AI	1	7.3	$0.751	1/2	66.0s
Total Tests 2 Wrong Tests 1 Total Cost $0.751 Response Time (avg) 66.0s
#20	Grok 4.5 low	X AI	1	6.5	$0.935	1/2	12.8s
Total Tests 2 Wrong Tests 1 Total Cost $0.935 Response Time (avg) 12.8s

←

1 2 3 4 5

Filter models