Agentic AI Atlas

II.

Benchmark JSON

benchmark:olympiad-bench

Structured · live

OlympiadBench json

Inspect the normalized record payload exactly as the atlas UI reads it.

File · benchmarks/benchmarks/benchmarks-knowledge.yamlCluster · benchmarks

Record JSON

{
  "id": "benchmark:olympiad-bench",
  "_kind": "Benchmark",
  "_file": "benchmarks/benchmarks/benchmarks-knowledge.yaml",
  "_cluster": "benchmarks",
  "attributes": {
    "displayName": "OlympiadBench",
    "homepageUrl": "https://github.com/OpenBMB/OlympiadBench",
    "kind": "model-only",
    "targetsKind": "ModelVersion",
    "description": "OlympiadBench (OpenBMB) is a bilingual (English/Chinese) multimodal\nbenchmark of 8,476 olympiad-level math and physics problems, with\nexpert-annotated step-by-step solutions, used to probe the limits\nof frontier-model reasoning.\n"
  },
  "outgoingEdges": [
    {
      "from": "benchmark:olympiad-bench",
      "to": "skill-area:mathematical-reasoning",
      "kind": "covers",
      "attributes": {}
    },
    {
      "from": "benchmark:olympiad-bench",
      "to": "domain:mathematics",
      "kind": "applies_to",
      "attributes": {}
    },
    {
      "from": "benchmark:olympiad-bench",
      "to": "domain:physics",
      "kind": "applies_to",
      "attributes": {}
    }
  ],
  "incomingEdges": []
}