Instructions following Model Ranking

See which AI models perform best on Instructions following, which ones stay reliable, and where the biggest gaps appear. Sort by: Tests Correct ↓.

Models Shown

Average Instructions following Score

0.8

Best Model

Gemini 3.6 Flash 1.0

Failure Reasons

With failure reason Wrong answer118 With failure reason Did not follow instructions82 With failure reason API error25 With failure reason Extra formatting8 With failure reason Timed out5 With failure reason No answer2

461/461

Rank	Model	Company	Instructions following Score	Score	Total Cost	Tests Correct	Response Time (avg)
#169	Gemma 4 26B A4B medium	Google	1.0	0.7	$0.101	2/2	17.5s
Total Tests 2 Wrong Tests 0 Total Cost $0.101 Response Time (avg) 17.5s
#170	GLM 5.3 Flash high	Z.ai	1.0	0.7	$0.056	2/2	2.06s
Total Tests 2 Wrong Tests 0 Total Cost $0.056 Response Time (avg) 2.06s
#173	Gemini 2.5 Flash medium	Google	1.0	0.7	$0.680	2/2	2.62s
Total Tests 2 Wrong Tests 0 Total Cost $0.680 Response Time (avg) 2.62s
#175	Qwen3.6 Plus medium	Qwen	1.0	0.7	$0.554	2/2	7.54s
Total Tests 2 Wrong Tests 0 Total Cost $0.554 Response Time (avg) 7.54s
#176	Gemini 3.5 Flash Lite medium	Google	1.0	0.7	$0.440	2/2	1.72s
Total Tests 2 Wrong Tests 0 Total Cost $0.440 Response Time (avg) 1.72s
#178	Qwen3.5 Plus 2026-02-15 medium	Qwen	1.0	0.7	$0.567	2/2	31.9s
Total Tests 2 Wrong Tests 0 Total Cost $0.567 Response Time (avg) 31.9s
#179	Qwen3.8 27B medium	Qwen	1.0	0.7	~$0.073	2/2	2.54s
Total Tests 2 Wrong Tests 0 Total Cost ~$0.073 Response Time (avg) 2.54s
#180	Gemini 3.8 Flash low	Google	1.0	0.7	$0.323	2/2	2.79s
Total Tests 2 Wrong Tests 0 Total Cost $0.323 Response Time (avg) 2.79s
#181	Claude Opus 4.8 low	Anthropic	1.0	0.7	$3.787	2/2	2.78s
Total Tests 2 Wrong Tests 0 Total Cost $3.787 Response Time (avg) 2.78s
#182	Qwen3.7 Flash high	Qwen	1.0	0.7	$0.063	2/2	9.41s
Total Tests 2 Wrong Tests 0 Total Cost $0.063 Response Time (avg) 9.41s
#184	GLM 5 medium	Z.ai	1.0	0.7	$0.479	2/2	7.25s
Total Tests 2 Wrong Tests 0 Total Cost $0.479 Response Time (avg) 7.25s
#185	Step 3.7 Flash low	Stepfun	1.0	0.7	$0.449	2/2	1.58s
Total Tests 2 Wrong Tests 0 Total Cost $0.449 Response Time (avg) 1.58s
#186	Qwen3.6 27B medium	Qwen	1.0	0.7	$1.021	2/2	38.0s
Total Tests 2 Wrong Tests 0 Total Cost $1.021 Response Time (avg) 38.0s
#187	Claude Opus 5 none	Anthropic	1.0	0.7	$2.766	2/2	2.81s
Total Tests 2 Wrong Tests 0 Total Cost $2.766 Response Time (avg) 2.81s
#188	Qwen3.8 Flash Next xhigh	Qwen	1.0	0.7	~$0.195	2/2	3.72s
Total Tests 2 Wrong Tests 0 Total Cost ~$0.195 Response Time (avg) 3.72s

←

1 9 10 11 31

→

Instructions following Ranking

Filter models

Top Models by Instructions following Score

Instructions following Score vs Total Cost

Top Models by Response Time (avg)

Instructions following Ranking

Filter models

Top Models by Instructions following Score

Instructions following Score vs Total CostEstimated GPU electricity cost only. It excludes the rest of the computer and hardware ownership. NVIDIA GeForce RTX 3090: 0.7349 h × 0.300 kW × €0.2896/kWh × 1.138 USD/EUR = ~$0.072660.

Top Models by Response Time (avg)

Instructions following Score vs Total Cost