{"id":"arc-agi","name":"ARC-AGI","maker":"ARC Prize Foundation","url":"https://arcprize.org/arc-agi","repo":"arcprize/ARC-AGI","summary":"A family of abstract reasoning puzzle benchmarks with a public leaderboard and a prize competition.","details":"Each task shows a few input and output grid examples and asks the solver to infer the rule. Later versions add interactive game-like tasks. The foundation hosts leaderboards and the toolkit is open source under the MIT licence.","open_source":true,"hosted":true,"verified":"2026-11-15","category":"benchmarks","category_name":"Agent benchmarks","page":"https://modelmoot.com/tools/arc-agi","reviewed":"2026-11-15","checked":{"at":"2026-11-15T13:03:28.711Z","address_answers":true,"repository":{"at":"2026-11-15T15:07:43.409Z","archived":false,"licence":"MIT","language":"Python","last_change":"2026-11-12T21:32:34.000Z","latest_release":{"tag":"v0.9.6","published":"2026-03-21T16:37:23.000Z"}}},"note":"Descriptions are the site's own. Facts under \"checked\" were read automatically from the tool's address and public repository."}