{"paper":{"title":"OpenAI Gym","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A toolkit supplies benchmark problems for reinforcement learning through a shared interface along with a website for comparing algorithm results.","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Greg Brockman, Jie Tang, John Schulman, Jonas Schneider, Ludwig Pettersson, Vicki Cheung, Wojciech Zaremba","submitted_at":"2016-06-05T17:54:48Z","abstract_excerpt":"OpenAI Gym is a toolkit for reinforcement learning research. It includes a growing collection of benchmark problems that expose a common interface, and a website where people can share their results and compare the performance of algorithms. This whitepaper discusses the components of OpenAI Gym and the design decisions that went into the software."},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"OpenAI Gym is a toolkit for reinforcement learning research. It includes a growing collection of benchmark problems that expose a common interface, and a website where people can share their results and compare the performance of algorithms.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That a common interface plus public result sharing will be sufficient to drive meaningful progress and fair comparisons, without addressing whether the chosen environments adequately represent real-world complexity or diversity of tasks.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"OpenAI Gym introduces a common interface for reinforcement learning environments and a results-sharing website to enable consistent algorithm comparisons.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A toolkit supplies benchmark problems for reinforcement learning through a shared interface along with a website for comparing algorithm results.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"b4e9a32b03e160a11aca312a59d6b682c72d3c4dbeae247e8c5f59635734a29b"},"source":{"id":"1606.01540","kind":"arxiv","version":1},"verdict":{"id":"0d263c97-fdee-4aa7-b588-f82c7144cd12","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-11T19:44:10.330352Z","strongest_claim":"OpenAI Gym is a toolkit for reinforcement learning research. It includes a growing collection of benchmark problems that expose a common interface, and a website where people can share their results and compare the performance of algorithms.","one_line_summary":"OpenAI Gym introduces a common interface for reinforcement learning environments and a results-sharing website to enable consistent algorithm comparisons.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That a common interface plus public result sharing will be sufficient to drive meaningful progress and fair comparisons, without addressing whether the chosen environments adequately represent real-world complexity or diversity of tasks.","pith_extraction_headline":"A toolkit supplies benchmark problems for reinforcement learning through a shared interface along with a website for comparing algorithm results."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/1606.01540/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":15,"sample":[{"doi":"","year":1995,"title":"Dynamic programming and optimal control","work_id":"b2ea3f9f-2ebe-41e2-8330-a140e7bf8c6b","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2015,"title":"V . Mnih, K. Kavukcuoglu, D. Silver, A. A. Rusu, J. Veness, M. G. Bellemare, A. Graves, M. Riedmiller, A. K. Fidjeland, G. Ostrovski, S. Petersen, Sadik Beattie, C., Antonoglou A., H. I., King, D. Kum","work_id":"b1509e5c-b3da-4ffa-9f09-45c396463a06","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2015,"title":"J. Schulman, S. Levine, P. Abbeel, M. I. Jordan, and P. Moritz. Trust region policy optimization. In ICML, pages 1889–1897, 2015","work_id":"041d632c-7466-47c8-ad82-9b9a540abe13","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2016,"title":"Asynchronous Methods for Deep Reinforcement Learning","work_id":"b54058f9-0b3b-46ae-a9a0-f22748625e2a","ref_index":4,"cited_arxiv_id":"1602.01783","is_internal_anchor":false},{"doi":"","year":2013,"title":"M. G. Bellemare, Y . Naddaf, J. Veness, and M. Bowling. The Arcade Learning Environment: An evaluation platform for general agents. J. Artif. Intell. Res., 47:253–279, 2013","work_id":"e9360171-a564-4345-b474-6d1665449d8d","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":15,"snapshot_sha256":"7cb6aa4d9ea2bb97016e0b2b108bc06c172f7da556469e1c6ace24c4574aa3d2","internal_anchors":1},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}