Anti-AI Tricks Model Ranking

AI BENCHY Category

See which AI models perform best on Anti-AI Tricks, which ones stay reliable, and where the biggest gaps appear. Sort by: Metric ↑.

Models Shown

Average Anti-AI Tricks Score

6.7

Best Model

Failure Reasons

With failure reason Wrong answer165 With failure reason Did not follow instructions17 With failure reason Extra formatting11 With failure reason Timed out4 With failure reason No answer1

Rank	Model	Company	Anti-AI Tricks Score	Score	Tests Correct	Response Time (avg)
#8	Qwen3.5 Plus 2026-02-15 medium	Qwen	8.2	8.5	3/4	45.8s
#47	Grok 4.20 medium	X AI	8.2	7.0	3/4	3.36s
#3	Claude Opus 4.7 medium	Anthropic	8.3	9.2	3/4	1.85s
#4	Claude Opus 4.7 none	Anthropic	8.3	9.2	3/4	2.12s
#6	Seed-2.0-Lite medium	Bytedance Seed	8.3	8.6	3/4	18.0s
#16	GPT-5.4 medium	OpenAI	8.3	8.2	3/4	4.11s
#21	Gemini 3 Flash Preview none	Google	8.3	8.1	3/4	1.25s
#22	Gemini 3.1 Flash Lite Preview low	Google	8.3	8.1	3/4	2.12s
#38	GPT-5.4 Nano medium	OpenAI	8.3	7.6	3/4	4.52s
#60	Gemma 4 26B A4B none	Google	8.3	6.2	3/4	1.28s
#27	DeepSeek V3.2 medium	DeepSeek	8.4	8.0	3/4	30.7s
#15	Gemini 2.5 Flash medium	Google	8.4	8.2	3/4	6.30s
#44	GPT-5.4 Mini medium	OpenAI	8.6	7.3	3/4	4.05s
#7	GPT-5.3-Codex medium	OpenAI	8.7	8.6	3/4	4.16s
#10	Qwen3.5-27B medium	Qwen	8.7	8.4	3/4	19.8s

←

1 2 3 4 5 6 7

→

Anti-AI Tricks Ranking