Did not follow instructions Failure Ranking

See which AI models run into Did not follow instructions most often, so you can spot reliability risks before choosing one. Sort by: Score ↑.

Models Shown

Total Failures

522

Most Affected Model

Categories

In category Puzzle Solving166 In category General Intelligence142 In category Instructions following77 In category Anti-AI Tricks52 In category Coding50 In category Tool Calling13 In category Agentic10 In category Data parsing and extraction8 In category Domain specific3 In category Combined1

293/293

Rank	Model	Company	Did not follow instructions Count	Score	Total Cost	Tests Correct	Response Time (avg)
#39	Claude Haiku 5.5 xhigh	Anthropic	1	0.9	$0.140	18/23	16.6s
Total Tests 23 Wrong Tests 5 Total Cost $0.140 Response Time (avg) 16.6s
#38	GPT-5.3-Codex medium	OpenAI	2	0.9	$1.318	17/23	18.9s
Total Tests 23 Wrong Tests 6 Total Cost $1.318 Response Time (avg) 18.9s
#37	Qwen3.8 Max (0902) high	Qwen	1	0.9	$3.306	18/23	186.1s
Total Tests 23 Wrong Tests 5 Total Cost $3.306 Response Time (avg) 186.1s
#36	DeepSeek V4.1 Flash 1017 low	DeepSeek	1	0.9	$0.355	16/23	32.5s
Total Tests 23 Wrong Tests 7 Total Cost $0.355 Response Time (avg) 32.5s
#34	Claude Sonnet 5.5 max	Anthropic	2	0.9	$8.557	18/23	84.3s
Total Tests 23 Wrong Tests 5 Total Cost $8.557 Response Time (avg) 84.3s
#30	Muse Spark Open high	Meta	1	0.9	$1.782	17/23	152.0s
Total Tests 23 Wrong Tests 6 Total Cost $1.782 Response Time (avg) 152.0s
#25	Claude Opus 5.5 high	Anthropic	1	0.9	$2.133	19/23	27.6s
Total Tests 23 Wrong Tests 4 Total Cost $2.133 Response Time (avg) 27.6s
#24	Claude Opus 5.5 low	Anthropic	1	0.9	$1.733	19/23	25.7s
Total Tests 23 Wrong Tests 4 Total Cost $1.733 Response Time (avg) 25.7s

Did not follow instructions Failures