{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:KQ3KH2EUT2DCCAHZ6FTOZNVZI2","short_pith_number":"pith:KQ3KH2EU","schema_version":"1.0","canonical_sha256":"5436a3e8949e862100f9f166ecb6b94699bef612eebd8a025c2452e9a6a41bd3","source":{"kind":"arxiv","id":"2504.03160","version":4},"attestation_state":"computed","paper":{"title":"DeepResearcher: Scaling Deep Research via Reinforcement Learning in Real-world Environments","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"End-to-end RL training on the open web lets LLM agents outperform prompt and RAG baselines by up to 28.9 points while developing planning and self-reflection.","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Dayuan Fu, Lyumanshan Ye, Pengfei Liu, Pengrui Lu, Xiangkun Hu, Xiaojie Cai, Yuxiang Zheng","submitted_at":"2025-04-04T04:41:28Z","abstract_excerpt":"Large Language Models (LLMs) equipped with web search capabilities have demonstrated impressive potential for deep research tasks. However, current approaches predominantly rely on either manually engineered prompts (prompt engineering-based) with brittle performance or reinforcement learning within controlled Retrieval-Augmented Generation (RAG) environments (RAG-based) that fail to capture the complexities of real-world interaction. In this paper, we introduce DeepResearcher, the first comprehensive framework for end-to-end training of LLM-based deep research agents through scaling reinforce"},"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":true,"formal_links_present":false},"canonical_record":{"source":{"id":"2504.03160","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-04-04T04:41:28Z","cross_cats_sorted":["cs.CL","cs.LG"],"title_canon_sha256":"7996dd2f6a35c010b6913abfcca9e480138c0124154c22fbbe4a9e4b99e57ace","abstract_canon_sha256":"5a6ac865085b2664dc78e29bc353e509efef8d2704b7e734cf0b98455ba9cff6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:46.763010Z","signature_b64":"lv8xdYESiWsZZXJtTa1rt7L3PS/5sdamd2Mr2kWXZTywODKu62bgEpAI78CB8ydAkcUbM0AXcICXQ0RbsoMjAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5436a3e8949e862100f9f166ecb6b94699bef612eebd8a025c2452e9a6a41bd3","last_reissued_at":"2026-05-17T23:38:46.762485Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:46.762485Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DeepResearcher: Scaling Deep Research via Reinforcement Learning in Real-world Environments","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"End-to-end RL training on the open web lets LLM agents outperform prompt and RAG baselines by up to 28.9 points while developing planning and self-reflection.","cross_cats":["cs.CL","cs.LG"],"primary_cat":"cs.AI","authors_text":"Dayuan Fu, Lyumanshan Ye, Pengfei Liu, Pengrui Lu, Xiangkun Hu, Xiaojie Cai, Yuxiang Zheng","submitted_at":"2025-04-04T04:41:28Z","abstract_excerpt":"Large Language Models (LLMs) equipped with web search capabilities have demonstrated impressive potential for deep research tasks. However, current approaches predominantly rely on either manually engineered prompts (prompt engineering-based) with brittle performance or reinforcement learning within controlled Retrieval-Augmented Generation (RAG) environments (RAG-based) that fail to capture the complexities of real-world interaction. In this paper, we introduce DeepResearcher, the first comprehensive framework for end-to-end training of LLM-based deep research agents through scaling reinforce"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"DeepResearcher achieves substantial improvements of up to 28.9 points over prompt engineering-based baselines and up to 7.2 points over RAG-based RL agents, with emergent cognitive behaviors including planning, cross-validation, self-reflection, and honesty.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the multi-agent browsing architecture can reliably extract information from arbitrary real-world webpage structures at scale without introducing systematic biases or instability that would undermine the reported performance gains.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"End-to-end RL in authentic web environments produces LLM research agents that outperform prompt-engineering and RAG-based baselines by up to 28.9 and 7.2 points respectively while exhibiting emergent planning, cross-validation, and self-reflection.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"End-to-end RL training on the open web lets LLM agents outperform prompt and RAG baselines by up to 28.9 points while developing planning and self-reflection.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"eea7bbcef4bd1c25a2bd03aed38de86b097ddc4e7746d2e24b0d8f87cf649252"},"source":{"id":"2504.03160","kind":"arxiv","version":4},"verdict":{"id":"e8c3be02-a679-4ccc-b3f5-bf3cc6a75514","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T19:55:39.109853Z","strongest_claim":"DeepResearcher achieves substantial improvements of up to 28.9 points over prompt engineering-based baselines and up to 7.2 points over RAG-based RL agents, with emergent cognitive behaviors including planning, cross-validation, self-reflection, and honesty.","one_line_summary":"End-to-end RL in authentic web environments produces LLM research agents that outperform prompt-engineering and RAG-based baselines by up to 28.9 and 7.2 points respectively while exhibiting emergent planning, cross-validation, and self-reflection.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the multi-agent browsing architecture can reliably extract information from arbitrary real-world webpage structures at scale without introducing systematic biases or instability that would undermine the reported performance gains.","pith_extraction_headline":"End-to-end RL training on the open web lets LLM agents outperform prompt and RAG baselines by up to 28.9 points while developing planning and self-reflection."},"references":{"count":18,"sample":[{"doi":"","year":2025,"title":"Qwen, :, An Yang, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chengyuan Li, Dayiheng Liu, Fei Huang, Haoran Wei, Huan Lin, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi ","work_id":"e84e2134-b769-4027-8225-600d73ff84a9","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Timo Schick, Jane Dwivedi-Yu, Roberto Dess`ı, Roberta Raileanu, Maria Lomeli, Eric Hambro, Luke Zettle- moyer, Nicola Cancedda, and Thomas Scialom. 2023. Toolformer: Language models can teach themselv","work_id":"6fd18ea3-e439-4393-b21e-7dbddf8d3dd4","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"R1-Searcher: Incentivizing the Search Capability in LLMs via Reinforcement Learning","work_id":"f5ed73d2-f2ff-4cf6-853d-3586333e44ef","ref_index":3,"cited_arxiv_id":"2503.05592","is_internal_anchor":true},{"doi":"","year":2025,"title":"Kimi k1.5: Scaling Reinforcement Learning with LLMs","work_id":"bff96ab1-bd6a-4585-be23-74fdb51969c7","ref_index":4,"cited_arxiv_id":"2501.12599","is_internal_anchor":true},{"doi":"","year":2022,"title":"Harsh Trivedi, Niranjan Balasubramanian, Tushar Khot, and Ashish Sabharwal. 2022. MuSiQue: Multihop questions via single-hop question composition. Transactions of the Association for Computational Lin","work_id":"9a1f5b9e-1481-4ef8-9826-dda294612579","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":18,"snapshot_sha256":"7eb220dfde26546444931055c723153fc67b25c640face9c0c5a4787d50e602a","internal_anchors":2},"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":"2504.03160","created_at":"2026-05-17T23:38:46.762581+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.03160v4","created_at":"2026-05-17T23:38:46.762581+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.03160","created_at":"2026-05-17T23:38:46.762581+00:00"},{"alias_kind":"pith_short_12","alias_value":"KQ3KH2EUT2DC","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"KQ3KH2EUT2DCCAHZ","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"KQ3KH2EU","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":31,"internal_anchor_count":31,"sample":[{"citing_arxiv_id":"2505.04588","citing_title":"ZeroSearch: Incentivize the Search Capability of LLMs without Searching","ref_index":48,"is_internal_anchor":true},{"citing_arxiv_id":"2601.06943","citing_title":"Watching, Reasoning, and Searching: A Video Deep Research Benchmark on Open Web for Agentic Video Reasoning","ref_index":18,"is_internal_anchor":true},{"citing_arxiv_id":"2605.00505","citing_title":"LLM-Oriented Information Retrieval: A Denoising-First Perspective","ref_index":242,"is_internal_anchor":true},{"citing_arxiv_id":"2605.15706","citing_title":"Differentiable Mixture-of-Agents Incentivizes Swarm Intelligence of Large Language Models","ref_index":72,"is_internal_anchor":true},{"citing_arxiv_id":"2605.20128","citing_title":"MixRea: Benchmarking Explicit-Implicit Reasoning in Large Language Models","ref_index":18,"is_internal_anchor":true},{"citing_arxiv_id":"2509.02547","citing_title":"The Landscape of Agentic Reinforcement Learning for LLMs: A Survey","ref_index":281,"is_internal_anchor":true},{"citing_arxiv_id":"2510.00568","citing_title":"ReSeek: A Self-Correcting Framework for Search Agents with Instructive Rewards","ref_index":24,"is_internal_anchor":true},{"citing_arxiv_id":"2510.00861","citing_title":"Erase to Improve: Erasable Reinforcement Learning for Search-Augmented LLMs","ref_index":43,"is_internal_anchor":true},{"citing_arxiv_id":"2510.19225","citing_title":"RLBoost: Harvesting Preemptible Resources for Cost-Efficient Reinforcement Learning on LLMs","ref_index":50,"is_internal_anchor":true},{"citing_arxiv_id":"2511.11793","citing_title":"MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scaling","ref_index":19,"is_internal_anchor":true},{"citing_arxiv_id":"2505.04588","citing_title":"ZeroSearch: Incentivize the Search Capability of LLMs without Searching","ref_index":48,"is_internal_anchor":true},{"citing_arxiv_id":"2507.02592","citing_title":"WebSailor: Navigating Super-human Reasoning for Web Agent","ref_index":34,"is_internal_anchor":true},{"citing_arxiv_id":"2601.12538","citing_title":"Agentic Reasoning for Large Language Models","ref_index":261,"is_internal_anchor":true},{"citing_arxiv_id":"2504.21776","citing_title":"WebThinker: Empowering Large Reasoning Models with Deep Research Capability","ref_index":69,"is_internal_anchor":true},{"citing_arxiv_id":"2506.20670","citing_title":"MMSearch-R1: Incentivizing LMMs to Search","ref_index":72,"is_internal_anchor":true},{"citing_arxiv_id":"2506.11763","citing_title":"DeepResearch Bench: A Comprehensive Benchmark for Deep Research Agents","ref_index":37,"is_internal_anchor":true},{"citing_arxiv_id":"2511.05271","citing_title":"DeepEyesV2: Toward Agentic Multimodal Model","ref_index":63,"is_internal_anchor":true},{"citing_arxiv_id":"2508.05748","citing_title":"WebWatcher: Breaking New Frontier of Vision-Language Deep Research Agent","ref_index":30,"is_internal_anchor":true},{"citing_arxiv_id":"2506.15841","citing_title":"MEM1: Learning to Synergize Memory and Reasoning for Efficient Long-Horizon Agents","ref_index":69,"is_internal_anchor":true},{"citing_arxiv_id":"2504.13958","citing_title":"ToolRL: Reward is All Tool Learning Needs","ref_index":45,"is_internal_anchor":true},{"citing_arxiv_id":"2605.07202","citing_title":"Towards Autonomous Business Intelligence via Data-to-Insight Discovery Agent","ref_index":5,"is_internal_anchor":true},{"citing_arxiv_id":"2605.10530","citing_title":"Personalized Deep Research: A User-Centric Framework, Dataset, and Hybrid Evaluation for Knowledge Discovery","ref_index":53,"is_internal_anchor":true},{"citing_arxiv_id":"2605.00505","citing_title":"LLM-Oriented Information Retrieval: A Denoising-First Perspective","ref_index":232,"is_internal_anchor":true},{"citing_arxiv_id":"2604.20146","citing_title":"SAKE: Self-aware Knowledge Exploitation-Exploration for Grounded Multimodal Named Entity Recognition","ref_index":51,"is_internal_anchor":true},{"citing_arxiv_id":"2604.20486","citing_title":"ProMMSearchAgent: A Generalizable Multimodal Search Agent Trained with Process-Oriented Rewards","ref_index":33,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KQ3KH2EUT2DCCAHZ6FTOZNVZI2","json":"https://pith.science/pith/KQ3KH2EUT2DCCAHZ6FTOZNVZI2.json","graph_json":"https://pith.science/api/pith-number/KQ3KH2EUT2DCCAHZ6FTOZNVZI2/graph.json","events_json":"https://pith.science/api/pith-number/KQ3KH2EUT2DCCAHZ6FTOZNVZI2/events.json","paper":"https://pith.science/paper/KQ3KH2EU"},"agent_actions":{"view_html":"https://pith.science/pith/KQ3KH2EUT2DCCAHZ6FTOZNVZI2","download_json":"https://pith.science/pith/KQ3KH2EUT2DCCAHZ6FTOZNVZI2.json","view_paper":"https://pith.science/paper/KQ3KH2EU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.03160&json=true","fetch_graph":"https://pith.science/api/pith-number/KQ3KH2EUT2DCCAHZ6FTOZNVZI2/graph.json","fetch_events":"https://pith.science/api/pith-number/KQ3KH2EUT2DCCAHZ6FTOZNVZI2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KQ3KH2EUT2DCCAHZ6FTOZNVZI2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KQ3KH2EUT2DCCAHZ6FTOZNVZI2/action/storage_attestation","attest_author":"https://pith.science/pith/KQ3KH2EUT2DCCAHZ6FTOZNVZI2/action/author_attestation","sign_citation":"https://pith.science/pith/KQ3KH2EUT2DCCAHZ6FTOZNVZI2/action/citation_signature","submit_replication":"https://pith.science/pith/KQ3KH2EUT2DCCAHZ6FTOZNVZI2/action/replication_record"}},"created_at":"2026-05-17T23:38:46.762581+00:00","updated_at":"2026-05-17T23:38:46.762581+00:00"}