Puzzle Solving Model Ranking

AI BENCHY Category

See which AI models perform best on Puzzle Solving, which ones stay reliable, and where the biggest gaps appear. Sort by: Score ↓.

Models Shown

Average Puzzle Solving Score

6.4

Best Model

Failure Reasons

With failure reason Wrong answer85 With failure reason Did not follow instructions75 With failure reason Timed out6 With failure reason Extra formatting2 With failure reason API error1

Rank	Model	Company	Puzzle Solving Score	Score	Tests Correct	Response Time (avg)
#1	Gemini 3 Flash Preview medium	Google	10.0	10.0	3/3	4.43s
#2	Gemini 3.1 Pro Preview medium	Google	10.0	9.6	3/3	7.15s
#3	Claude Opus 4.7 medium	Anthropic	10.0	9.2	3/3	2.51s
#4	Claude Opus 4.7 none	Anthropic	10.0	9.2	3/3	2.58s
#5	Gemini 3 Flash Preview low	Google	10.0	8.8	3/3	6.11s
#6	Seed-2.0-Lite medium	Bytedance Seed	9.0	8.6	2/3	11.0s
#7	GPT-5.3-Codex medium	OpenAI	9.0	8.6	2/3	5.12s
#8	Qwen3.5 Plus 2026-02-15 medium	Qwen	10.0	8.5	3/3	34.6s
#9	Qwen3.6 Plus Preview medium	Qwen	10.0	8.5	3/3	6.11s
#10	Qwen3.5-27B medium	Qwen	8.2	8.4	2/3	64.6s
#11	Gemini 3.1 Flash Lite Preview high	Google	7.7	8.4	2/3	46.3s
#12	Gemini 3 PRO Preview medium	Google	10.0	8.4	3/3	3.91s
#13	GLM 5 medium	Z.ai	10.0	8.4	3/3	15.6s
#14	Gemma 4 31B medium	Google	8.8	8.3	2/3	27.6s
#15	Gemini 2.5 Flash medium	Google	7.7	8.2	2/3	3.94s

1 2 3 4 5 6 7

→

Puzzle Solving Ranking