Share your thoughts, 1 month free Claude Pro on usSee more

Function Calling on ToolBench Average

75.95Pass Rate

ParaTool

Updated 1mo ago

Evaluation Results

Method	Links
ParaTool 2026.05		75.95	80.75
ParaTool 2026.05		69.57	71.82
ToolLLaMA 2026.05		68.31	70.18
Parameter+Reinvoke 2026.05		68.28	71.28
Context+Docs & Examples 2026.05		66.96	69.08
Context+Docs & Examples 2026.05		65.67	63.58
ToolLLaMA 2026.05		65.41	65.5
Parameter+Reinvoke 2026.05		63.78	65.13
Claude-3.5-Sonnet + TAFC 2026.01		60.3	72.2
Parameter+ToolRetriever 2026.05		60.26	62.76
Parameter+ToolRetriever 2026.05		60.05	60.6
GPT-4o + TAFC 2026.01		59.2	71.4
Claude-3.5-Sonnet 2026.01		58.4	70.5
Context+Docs 2026.05		58.06	-
GPT-4o 2026.01		57.5	69.4
Parameter+Embsim 2026.05		57.04	61.91
Parameter+Embsim 2026.05		56.21	57.59
Context+Docs 2026.05		56.13	-
Qwen2.5-72B + TAFC 2026.01		49	57.1
Qwen2.5-72B 2026.01		46.9	55
Llama-3.1-70B + TAFC 2026.01		46.4	54.4
Llama-3.1-70B 2026.01		44.2	52
Qwen2.5-32B + TAFC 2026.01		42.7	50.8
Qwen2.5-32B 2026.01		40.5	48.3
Global Parameterization 2026.05		39.33	38.19
Global Parameterization 2026.05		35.2	32.56
Qwen2.5-7B + TAFC 2026.01		31.2	38
Llama-3.1-8B + TAFC 2026.01		29.7	36.3
Qwen2.5-7B 2026.01		28.7	34.9
Llama-3.1-8B 2026.01		27.3	33.4