{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:5NIZO4ZJOXNHUHSGP2NCPWXVZX","short_pith_number":"pith:5NIZO4ZJ","schema_version":"1.0","canonical_sha256":"eb5197732975da7a1e467e9a27daf5cdfe9f718797e31ee233900eac0515a6b6","source":{"kind":"arxiv","id":"2409.08264","version":2},"attestation_state":"computed","paper":{"title":"Windows Agent Arena: Evaluating Multi-Modal OS Agents at Scale","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Arthur Bucker, Dan Zhao, Dillon Dupont, Francesco Bonacci, Justin Wagle, Kazuhito Koishida, Lawrence Jang, Rogerio Bonatti, Sara Abdali, Yadong Lu, Yinheng Li, Zack Hui","submitted_at":"2024-09-12T17:56:43Z","abstract_excerpt":"Large language models (LLMs) show remarkable potential to act as computer agents, enhancing human productivity and software accessibility in multi-modal tasks that require planning and reasoning. However, measuring agent performance in realistic environments remains a challenge since: (i) most benchmarks are limited to specific modalities or domains (e.g. text-only, web navigation, Q&A, coding) and (ii) full benchmark evaluations are slow (on order of magnitude of days) given the multi-step sequential nature of tasks. To address these challenges, we introduce the Windows Agent Arena: a reprodu"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2409.08264","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2024-09-12T17:56:43Z","cross_cats_sorted":[],"title_canon_sha256":"9828db3fda786215059d5fc71ae832b2809152f2891279174aea3d676fe81afe","abstract_canon_sha256":"6134d75817e9f57b1f4ac23dc1b54e03a7f062e5212c5582b7d3da6359daedf7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:07:03.926711Z","signature_b64":"I1hplO83eXVoPDhBcwSoWPkxpnIkgBsiO4NcqJsIA3tGzDAwOic44OKBpTjkzgrsWjRQ6y31SPvIzuncfmh4Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"eb5197732975da7a1e467e9a27daf5cdfe9f718797e31ee233900eac0515a6b6","last_reissued_at":"2026-07-05T09:07:03.926221Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:07:03.926221Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Windows Agent Arena: Evaluating Multi-Modal OS Agents at Scale","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Arthur Bucker, Dan Zhao, Dillon Dupont, Francesco Bonacci, Justin Wagle, Kazuhito Koishida, Lawrence Jang, Rogerio Bonatti, Sara Abdali, Yadong Lu, Yinheng Li, Zack Hui","submitted_at":"2024-09-12T17:56:43Z","abstract_excerpt":"Large language models (LLMs) show remarkable potential to act as computer agents, enhancing human productivity and software accessibility in multi-modal tasks that require planning and reasoning. However, measuring agent performance in realistic environments remains a challenge since: (i) most benchmarks are limited to specific modalities or domains (e.g. text-only, web navigation, Q&A, coding) and (ii) full benchmark evaluations are slow (on order of magnitude of days) given the multi-step sequential nature of tasks. To address these challenges, we introduce the Windows Agent Arena: a reprodu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.08264","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2409.08264/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2409.08264","created_at":"2026-07-05T09:07:03.926280+00:00"},{"alias_kind":"arxiv_version","alias_value":"2409.08264v2","created_at":"2026-07-05T09:07:03.926280+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.08264","created_at":"2026-07-05T09:07:03.926280+00:00"},{"alias_kind":"pith_short_12","alias_value":"5NIZO4ZJOXNH","created_at":"2026-07-05T09:07:03.926280+00:00"},{"alias_kind":"pith_short_16","alias_value":"5NIZO4ZJOXNHUHSG","created_at":"2026-07-05T09:07:03.926280+00:00"},{"alias_kind":"pith_short_8","alias_value":"5NIZO4ZJ","created_at":"2026-07-05T09:07:03.926280+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":32,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.23049","citing_title":"PhoneBuddy: Training Open Models for Agentic Phone Use","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17929","citing_title":"PreAct: Computer-Using Agents that Get Faster on Repeated Tasks","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17321","citing_title":"ProCUA-SFT Technical Report","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2606.12191","citing_title":"Agentic Environment Engineering for Large Language Models: A Survey of Environment Modeling, Synthesis, Evaluation, and Application","ref_index":55,"is_internal_anchor":false},{"citing_arxiv_id":"2606.12195","citing_title":"InternVideo3: Agentify Foundation Models with Multimodal Contextual Reasoning","ref_index":96,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01936","citing_title":"What to Format and How: A Benchmark and Workflow Approach for Document Formatting","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2605.24785","citing_title":"PANDO: Efficient Multimodal AI Agents via Online Skill Distillation","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.25160","citing_title":"ScaleWoB: Guiding GUI Agents with Coding Agents via Large-Scale Environmental Synthesis","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28715","citing_title":"SEATauBench: Adapting Tool-Agent-User Evaluation Into Low-Resource Southeast Asian Languages","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2605.25343","citing_title":"Toward Native Multimodal Modeling: A Roadmap","ref_index":134,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27761","citing_title":"AndroidDaily: A Verifiable Benchmark for Mobile GUI Agents on Real-World Closed-Source Applications","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2605.28775","citing_title":"Learn from Weaknesses: Automated Domain Specialization for Small Computer-Use Agents","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2605.29115","citing_title":"unix-ctf: Procedural Environments for Unix-Competence Reinforcement Learning","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2501.16150","citing_title":"A Comprehensive Survey of Agents for Computer Use: Foundations, Challenges, and Future Directions","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2605.16402","citing_title":"WinDeskGround: A Benchmark for Robust GUI Grounding in Complex Multi-Window Desktop Environments","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19769","citing_title":"OpenComputer: Verifiable Software Worlds for Computer-Use Agents","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2506.02387","citing_title":"VS-Bench: Evaluating VLMs for Strategic Abilities in Multi-Agent Environments","ref_index":10,"is_internal_anchor":false},{"citing_arxiv_id":"2504.14239","citing_title":"InfiGUI-R1: Advancing Multimodal GUI Agents from Reactive Actors to Deliberative Reasoners","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2509.21816","citing_title":"From Task to Tutorial: An Automated GUI Framework for Excel Tutorial Document and Video Creation","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2510.05307","citing_title":"When Should Users Check? Modeling Confirmation Frequency inMulti-Step Agentic AI Tasks","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27996","citing_title":"Exploring LLM Agent Designs and Interaction Modalities for Scientific Visualization","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2405.14573","citing_title":"AndroidWorld: A Dynamic Benchmarking Environment for Autonomous Agents","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12501","citing_title":"Covering Human Action Space for Computer Use: Data Synthesis and Benchmark","ref_index":35,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12481","citing_title":"ToolCUA: Towards Optimal GUI-Tool Path Orchestration for Computer Use Agents","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27955","citing_title":"GUI Agents with Reinforcement Learning: Toward Digital Inhabitants","ref_index":8,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5NIZO4ZJOXNHUHSGP2NCPWXVZX","json":"https://pith.science/pith/5NIZO4ZJOXNHUHSGP2NCPWXVZX.json","graph_json":"https://pith.science/api/pith-number/5NIZO4ZJOXNHUHSGP2NCPWXVZX/graph.json","events_json":"https://pith.science/api/pith-number/5NIZO4ZJOXNHUHSGP2NCPWXVZX/events.json","paper":"https://pith.science/paper/5NIZO4ZJ"},"agent_actions":{"view_html":"https://pith.science/pith/5NIZO4ZJOXNHUHSGP2NCPWXVZX","download_json":"https://pith.science/pith/5NIZO4ZJOXNHUHSGP2NCPWXVZX.json","view_paper":"https://pith.science/paper/5NIZO4ZJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2409.08264&json=true","fetch_graph":"https://pith.science/api/pith-number/5NIZO4ZJOXNHUHSGP2NCPWXVZX/graph.json","fetch_events":"https://pith.science/api/pith-number/5NIZO4ZJOXNHUHSGP2NCPWXVZX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5NIZO4ZJOXNHUHSGP2NCPWXVZX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5NIZO4ZJOXNHUHSGP2NCPWXVZX/action/storage_attestation","attest_author":"https://pith.science/pith/5NIZO4ZJOXNHUHSGP2NCPWXVZX/action/author_attestation","sign_citation":"https://pith.science/pith/5NIZO4ZJOXNHUHSGP2NCPWXVZX/action/citation_signature","submit_replication":"https://pith.science/pith/5NIZO4ZJOXNHUHSGP2NCPWXVZX/action/replication_record"}},"created_at":"2026-07-05T09:07:03.926280+00:00","updated_at":"2026-07-05T09:07:03.926280+00:00"}