Tool Calling x Wrong answer Ranking

See which AI models are most likely to hit Wrong answer on Tool Calling, so you can spot weak points faster.

Models Shown

Total Failures

Most Affected Model

Failure Reasons

API error25 Did not follow instructions12 Invalid tool call11 Wrong answer5 No answer2

Categories

Domain specific654 Anti-AI Tricks386 Coding361 Puzzle Solving275 Trivia244 Combined95 General Intelligence90 Instructions following82 Data parsing and extraction66 Tool Calling5

5/5

Rank	Model	Company	Wrong answer Count	Category Score	Total Cost	Tests Correct	Response Time (avg)
#69	Qwen3.8 27B low	Qwen	1	3.0	~$0.071	0/1	8.42s
Total Tests 1 Wrong Tests 1 Total Cost ~$0.071 Response Time (avg) 8.42s
#91	GPT-5.6 Terra low	OpenAI	1	4.7	$0.420	0/1	6.69s
Total Tests 1 Wrong Tests 1 Total Cost $0.420 Response Time (avg) 6.69s
#275	GLM 4.7 Flash none	Z.ai	1	2.8	$0.016	0/1	7.05s
Total Tests 1 Wrong Tests 1 Total Cost $0.016 Response Time (avg) 7.05s
#285	Laguna S 2.1 none	Poolside	1	4.7	$0.022	0/1	1.45s
Total Tests 1 Wrong Tests 1 Total Cost $0.022 Response Time (avg) 1.45s
#303	Grok 4.1 Fast none	X AI	1	2.8	$0.008	0/1	5.51s
Total Tests 1 Wrong Tests 1 Total Cost $0.008 Response Time (avg) 5.51s

Filter models