{
  "id": "benchmark:toolbench",
  "_kind": "Benchmark",
  "_file": "benchmarks/benchmarks/benchmarks-tool-use.yaml",
  "_cluster": "benchmarks",
  "attributes": {
    "displayName": "ToolBench",
    "homepageUrl": "https://github.com/OpenBMB/ToolBench",
    "kind": "agent-platform",
    "targetsKind": "AgentVersion",
    "description": "ToolBench (OpenBMB) is a large-scale instruction-tuning and\nevaluation suite for LLM tool-use, built on 16,000+ real-world\nREST APIs from RapidAPI; the companion ToolEval harness scores\npass-rate and win-rate against a reference tool-using agent.\n"
  },
  "outgoingEdges": [
    {
      "from": "benchmark:toolbench",
      "to": "skill-area:tool-use",
      "kind": "covers",
      "attributes": {}
    },
    {
      "from": "benchmark:toolbench",
      "to": "skill-area:multi-turn-tool-use",
      "kind": "covers",
      "attributes": {}
    },
    {
      "from": "benchmark:toolbench",
      "to": "domain:software-engineering",
      "kind": "applies_to",
      "attributes": {}
    },
    {
      "from": "benchmark:toolbench",
      "to": "domain:api-development",
      "kind": "applies_to",
      "attributes": {}
    }
  ],
  "incomingEdges": []
}