Coding Model Ranking

AI BENCHY Category

See which AI models perform best on Coding, which ones stay reliable, and where the biggest gaps appear. Sort by: Response Time (avg) ↑.

Models Shown

Average Coding Score

5.7

Best Model

Failure Reasons

With failure reason Wrong answer230 With failure reason API error43 With failure reason Timed out23 With failure reason No answer18 With failure reason Did not follow instructions16 With failure reason Extra formatting12

188/188

Rank	Model	Company	Coding Score	Score	Total Cost	Tests Correct	Response Time (avg)
#55	Kimi K2.5 medium	Moonshot AI	6.1	7.5	$0.348	1/3	217.5s
Total Tests 3 Wrong Tests 2 Total Cost $0.348 Response Time (avg) 217.5s
#102	Gemma 4 31B medium	Google	4.3	6.3	$0.033	0/3	219.8s
Total Tests 3 Wrong Tests 3 Total Cost $0.033 Response Time (avg) 219.8s
#62	Seed-2.0-Mini medium	Bytedance Seed	5.5	7.4	$0.044	1/3	220.5s
Total Tests 3 Wrong Tests 2 Total Cost $0.044 Response Time (avg) 220.5s
#52	DeepSeek V4 Pro high	DeepSeek	6.1	7.6	$0.157	1/3	243.0s
Total Tests 3 Wrong Tests 2 Total Cost $0.157 Response Time (avg) 243.0s
#60	DeepSeek V3.2 medium	DeepSeek	6.0	7.5	$0.044	1/3	248.7s
Total Tests 3 Wrong Tests 2 Total Cost $0.044 Response Time (avg) 248.7s
#93	Step 3.5 Flash medium	Stepfun	2.4	6.6	$0.070	0/2	258.4s
Total Tests 2 Wrong Tests 2 Total Cost $0.070 Response Time (avg) 258.4s
#71	Gemma 4 26B A4B medium	Google	2.9	7.2	$0.045	0/3	272.5s
Total Tests 3 Wrong Tests 3 Total Cost $0.045 Response Time (avg) 272.5s
#123	North Mini Code medium	Cohere	4.5	5.8	$0.000	0/3	320.4s
Total Tests 3 Wrong Tests 3 Total Cost $0.000 Response Time (avg) 320.4s

Coding Ranking