{"id":"toolathlon","name":"Toolathlon","maker":"HKUST NLP Group and collaborators","url":"https://toolathlon.xyz","repo":"hkust-nlp/Toolathlon","summary":"A benchmark of long, realistic tasks that need many tools across real software applications.","details":"Tasks run against real applications through MCP servers and each has a checker that verifies the end state. The repository holds the environments and evaluation code.","open_source":true,"hosted":false,"speaks":["mcp"],"verified":"2026-11-15","category":"benchmarks","category_name":"Agent benchmarks","page":"https://modelmoot.com/tools/toolathlon","reviewed":"2026-11-15","checked":{"at":"2026-11-15T16:10:15.342Z","address_answers":true},"note":"Descriptions are the site's own. Facts under \"checked\" were read automatically from the tool's address and public repository."}