Tool-use behavioral profiling on Gray regime n=30 benchmark leaderboard · SOTA2 Research

Tool-use behavioral profiling on Gray regime n=30

93A Scoredeepseek-chat

Evaluation Results

Method	Links
deepseek-chat2026.04		93	7	0
deepseek-chat2026.04		93	7	0
llama3.1:latest2026.04		80	13	13
mistral:latest2026.04		77	3	20
mistral:latest2026.04		73	3	30
mistral:latest2026.04		67	0	33
qwen2.5:7b-instruct2026.04		63	30	13
qwen2.5:7b-instruct2026.04		57	40	10
qwen2.5:7b-instruct2026.04		43	50	7
deepseek-chat2026.04		33	67	0
llama3.1:latest2026.04		30	60	23
llama3.1:latest2026.04		17	87	10