iiRecord
Agentic AI Atlas · ToolBench
benchmark:toolbencha5c.ai
II.
Benchmark overview

benchmark:toolbench

Reference · live

ToolBench overview

ToolBench (OpenBMB) is a large-scale instruction-tuning and evaluation suite for LLM tool-use, built on 16,000+ real-world REST APIs from RapidAPI; the companion ToolEval harness scores pass-rate and win-rate against a reference tool-using agent.

BenchmarkOutgoing · 4Incoming · 1

Attributes

displayName
ToolBench
homepageUrl
kind
agent-platform
targetsKind
AgentVersion
description
ToolBench (OpenBMB) is a large-scale instruction-tuning and evaluation suite for LLM tool-use, built on 16,000+ real-world REST APIs from RapidAPI; the companion ToolEval harness scores pass-rate and win-rate against a reference tool-using agent.

Outgoing edges

applies_to2
covers2

Incoming edges

belongs_to_benchmark1