Instructions following x Wrong answer Ranking

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

Models Shown

Total Failures

111

Most Affected Model

Failure Reasons

Wrong answer111 Did not follow instructions73 API error25 Extra formatting8 Timed out5 No answer2

Categories

Domain specific878 Anti-AI Tricks502 Coding477 Puzzle Solving344 Trivia306 General Intelligence129 Agentic126 Combined126 Instructions following111 Data parsing and extraction93

111/111

Rank	Model	Company	Wrong answer Count	Category Score	Total Cost	Tests Correct	Response Time (avg)
#360	North Mini Code default	Cohere	1	0.7	$0.000	1/2	30.7s
Total Tests 2 Wrong Tests 1 Total Cost $0.000 Response Time (avg) 30.7s
#370	Owl Alpha medium	Openrouter	1	0.7	$0.000	1/2	10.2s
Total Tests 2 Wrong Tests 1 Total Cost $0.000 Response Time (avg) 10.2s
#372	Owl Alpha default	Openrouter	1	0.6	$0.000	1/2	2.63s
Total Tests 2 Wrong Tests 1 Total Cost $0.000 Response Time (avg) 2.63s
#395	Ling 3.0 Tiny default	Inclusionai	1	0.3	$0.000	0/2	2.44s
Total Tests 2 Wrong Tests 2 Total Cost $0.000 Response Time (avg) 2.44s
#406	Hunter Alpha default	OpenRouter	1	0.6	$0.000	1/2	2.82s
Total Tests 2 Wrong Tests 1 Total Cost $0.000 Response Time (avg) 2.82s
#422	Nemotron 3 Nano Omni 30b A3b Reasoning default	NVIDIA	1	0.5	$0.000	0/2	541ms
Total Tests 2 Wrong Tests 2 Total Cost $0.000 Response Time (avg) 541ms

Filter models