Agentic Model Ranking

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

Models Shown

Average Agentic Score

5.8

Best Model

Failure Reasons

With failure reason Wrong answer108 With failure reason Invalid tool call44 With failure reason API error32 With failure reason Did not follow instructions14 With failure reason Timed out10

380/380

Rank	Model	Company	Agentic Score	Score	Total Cost	Tests Correct	Response Time (avg)
#229	DeepSeek V4 Flash 0731 none	DeepSeek	10.0	6.2	$0.055	1/1	174.9s
Total Tests 1 Wrong Tests 0 Total Cost $0.055 Response Time (avg) 174.9s
#234	Inkling none	Thinkingmachines	10.0	6.1	$0.304	1/1	45.6s
Total Tests 1 Wrong Tests 0 Total Cost $0.304 Response Time (avg) 45.6s
#235	Granite 4.2 8B medium	IBM Granite	10.0	6.1	$0.029	1/1	149.5s
Total Tests 1 Wrong Tests 0 Total Cost $0.029 Response Time (avg) 149.5s
#253	Trinity Large Thinking high	Arcee AI	10.0	5.7	$0.645	1/1	45.0s
Total Tests 1 Wrong Tests 0 Total Cost $0.645 Response Time (avg) 45.0s
#378	Gemini 3 Flash Preview high	Google	10.0	1.8	$0.105	1/1	52.5s
Total Tests 1 Wrong Tests 0 Total Cost $0.105 Response Time (avg) 52.5s

Agentic Ranking