List your company

Tool Execution on ToolBench With Ground-Truth Tools

60.7SoPR (I1)ToolOmni
22.7432.59542.4552.305Apr 15, 2026

Evaluation results

MethodLinks
ToolOmniInteraction budget (turns)=62026.0460.750.952.554.752.7656.640.9850.11
GPT-3.5Interaction budget (turns)=62026.0456.647.854.6453.01————
ToolGenModel size (parameters)=7B, Interaction budget (turns)=62026.0444.242.93239.735.5845.2816.3932.42
ToolLlama-v2Interaction budget (turns)=62026.0443.645.839.342.938.6555.6632.7942.37
ToolGenModel size (parameters)=3B, Interaction budget (turns)=62026.0440.843.421.335.1739.2647.1714.7533.73
ToolLlama-v1Interaction budget (turns)=62026.0424.229.212.321.928.2242.4516.3929.02