{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:M2KQFQTZY6GYBGKAP6QP7DYZUW","short_pith_number":"pith:M2KQFQTZ","canonical_record":{"source":{"id":"2506.20670","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-06-25T17:59:42Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"8ecc7b053022d2d258e49fd9c9a94ad2866fd3f21cc2e0a15ea9923f28eb1f08","abstract_canon_sha256":"d87506b60c8c6cbcb204c48af3d47e2d19d61c1bbb8a14d3a703a876367c7425"},"schema_version":"1.0"},"canonical_sha256":"669502c279c78d8099407fa0ff8f19a5a7a416364ed7208ee54d159fe652127e","source":{"kind":"arxiv","id":"2506.20670","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2506.20670","created_at":"2026-05-17T23:38:47Z"},{"alias_kind":"arxiv_version","alias_value":"2506.20670v1","created_at":"2026-05-17T23:38:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.20670","created_at":"2026-05-17T23:38:47Z"},{"alias_kind":"pith_short_12","alias_value":"M2KQFQTZY6GY","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"M2KQFQTZY6GYBGKA","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"M2KQFQTZ","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:M2KQFQTZY6GYBGKAP6QP7DYZUW","target":"record","payload":{"canonical_record":{"source":{"id":"2506.20670","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-06-25T17:59:42Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"8ecc7b053022d2d258e49fd9c9a94ad2866fd3f21cc2e0a15ea9923f28eb1f08","abstract_canon_sha256":"d87506b60c8c6cbcb204c48af3d47e2d19d61c1bbb8a14d3a703a876367c7425"},"schema_version":"1.0"},"canonical_sha256":"669502c279c78d8099407fa0ff8f19a5a7a416364ed7208ee54d159fe652127e","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:38:47.444920Z","signature_b64":"qekVEaEoh4rFM9FM5LvrDkpfPnAlMrKGR9suWfXA4LF4aQ4eHd1qSYhOp4/Blj0nIWEuc3i2OTcH3zKKL3u8CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"669502c279c78d8099407fa0ff8f19a5a7a416364ed7208ee54d159fe652127e","last_reissued_at":"2026-05-17T23:38:47.444503Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:38:47.444503Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2506.20670","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:47Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"y6VJet/IE3ROIkmySA5+Uy9vPGzwvrcdx2oUDdGIIk+0Mfu8ddlv4Ina4rEbXJZR/FQwxkg6/PqDcRvVCjJ0Dw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-05T09:44:16.152715Z"},"content_sha256":"839be17c1a7c264f7a32fc413c22b3b36cc6138c3552fc6ecae4b151be147243","schema_version":"1.0","event_id":"sha256:839be17c1a7c264f7a32fc413c22b3b36cc6138c3552fc6ecae4b151be147243"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:M2KQFQTZY6GYBGKAP6QP7DYZUW","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"MMSearch-R1: Incentivizing LMMs to Search","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Reinforcement learning lets multimodal models search the internet only when needed","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Bo Li, Bo You, Jinming Wu, Wei Li, Yiding Liu, Zejun Ma, Zihao Deng, Ziwei Liu","submitted_at":"2025-06-25T17:59:42Z","abstract_excerpt":"Robust deployment of large multimodal models (LMMs) in real-world scenarios requires access to external knowledge sources, given the complexity and dynamic nature of real-world information. Existing approaches such as retrieval-augmented generation (RAG) and prompt engineered search agents rely on rigid pipelines, often leading to inefficient or excessive search behaviors. We present MMSearch-R1, the first end-to-end reinforcement learning framework that enables LMMs to perform on-demand, multi-turn search in real-world Internet environments. Our framework integrates both image and text search"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We present MMSearch-R1, the first end-to-end reinforcement learning framework that enables LMMs to perform on-demand, multi-turn search in real-world Internet environments... our model not only outperforms RAG-based baselines of the same model size, but also matches the performance of a larger RAG-based model while reducing search calls by over 30%.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The outcome-based reward combined with a search penalty, together with the curated search-balanced dataset, is sufficient to produce efficient on-demand search behavior that generalizes beyond the training distribution.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"MMSearch-R1 uses reinforcement learning to train multimodal models for on-demand multi-turn internet search with image and text tools, outperforming same-size RAG baselines and matching larger ones while cutting search calls by over 30%.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Reinforcement learning lets multimodal models search the internet only when needed","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"3447c2a81aad5ee45d9eec2cce236f98202eda62d8441b689c80c1450340ee79"},"source":{"id":"2506.20670","kind":"arxiv","version":1},"verdict":{"id":"18f9ea2e-53a7-4c65-865b-759ce7bedc8c","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T15:21:56.253302Z","strongest_claim":"We present MMSearch-R1, the first end-to-end reinforcement learning framework that enables LMMs to perform on-demand, multi-turn search in real-world Internet environments... our model not only outperforms RAG-based baselines of the same model size, but also matches the performance of a larger RAG-based model while reducing search calls by over 30%.","one_line_summary":"MMSearch-R1 uses reinforcement learning to train multimodal models for on-demand multi-turn internet search with image and text tools, outperforming same-size RAG baselines and matching larger ones while cutting search calls by over 30%.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The outcome-based reward combined with a search penalty, together with the curated search-balanced dataset, is sufficient to produce efficient on-demand search behavior that generalizes beyond the training distribution.","pith_extraction_headline":"Reinforcement learning lets multimodal models search the internet only when needed"},"references":{"count":94,"sample":[{"doi":"","year":2025,"title":"Open deep search: Democratizing search with open-source reasoning agents","work_id":"467e5b5b-a169-4539-8ddc-bd34e07b0e3a","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Claude 3.5 Sonnet","work_id":"45d58c07-b237-4bb8-ac8d-8c68f4784507","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Self-rag: Learn- ing to retrieve, generate, and critique through self-reflection","work_id":"cd9884d5-53dc-4ba6-98d6-a38b02cf052e","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Mint-1t: Scaling open-source multimodal data by 10x: A multimodal dataset with one trillion tokens","work_id":"9350a215-8bdd-4a04-a217-295413298f83","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Qwen2.5-VL Technical Report","work_id":"69dffacb-bfe8-442d-be86-48624c60426f","ref_index":5,"cited_arxiv_id":"2502.13923","is_internal_anchor":true}],"resolved_work":94,"snapshot_sha256":"2a80b22b2e8323d5b48d357cc8e67cc82540c731f2a5ec48516d7e6cffe9fca1","internal_anchors":26},"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"},"verdict_id":"18f9ea2e-53a7-4c65-865b-759ce7bedc8c"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:38:47Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Mtzfc6YSzBbmgXT8z+2GZGfO0gp001WJKzV8VtgzuJLbmMCVtVwXTA/xz6USTigQbCs/vFckTtGcr47nHz6VDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-05T09:44:16.153632Z"},"content_sha256":"f72e4d3b8c0be2bd3f472f32b8a82620a78675619a700dc04441c9a03635c536","schema_version":"1.0","event_id":"sha256:f72e4d3b8c0be2bd3f472f32b8a82620a78675619a700dc04441c9a03635c536"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/M2KQFQTZY6GYBGKAP6QP7DYZUW/bundle.json","state_url":"https://pith.science/pith/M2KQFQTZY6GYBGKAP6QP7DYZUW/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/M2KQFQTZY6GYBGKAP6QP7DYZUW/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-06-05T09:44:16Z","links":{"resolver":"https://pith.science/pith/M2KQFQTZY6GYBGKAP6QP7DYZUW","bundle":"https://pith.science/pith/M2KQFQTZY6GYBGKAP6QP7DYZUW/bundle.json","state":"https://pith.science/pith/M2KQFQTZY6GYBGKAP6QP7DYZUW/state.json","well_known_bundle":"https://pith.science/.well-known/pith/M2KQFQTZY6GYBGKAP6QP7DYZUW/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:M2KQFQTZY6GYBGKAP6QP7DYZUW","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"d87506b60c8c6cbcb204c48af3d47e2d19d61c1bbb8a14d3a703a876367c7425","cross_cats_sorted":["cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-06-25T17:59:42Z","title_canon_sha256":"8ecc7b053022d2d258e49fd9c9a94ad2866fd3f21cc2e0a15ea9923f28eb1f08"},"schema_version":"1.0","source":{"id":"2506.20670","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2506.20670","created_at":"2026-05-17T23:38:47Z"},{"alias_kind":"arxiv_version","alias_value":"2506.20670v1","created_at":"2026-05-17T23:38:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.20670","created_at":"2026-05-17T23:38:47Z"},{"alias_kind":"pith_short_12","alias_value":"M2KQFQTZY6GY","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"M2KQFQTZY6GYBGKA","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"M2KQFQTZ","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:f72e4d3b8c0be2bd3f472f32b8a82620a78675619a700dc04441c9a03635c536","target":"graph","created_at":"2026-05-17T23:38:47Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"We present MMSearch-R1, the first end-to-end reinforcement learning framework that enables LMMs to perform on-demand, multi-turn search in real-world Internet environments... our model not only outperforms RAG-based baselines of the same model size, but also matches the performance of a larger RAG-based model while reducing search calls by over 30%."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The outcome-based reward combined with a search penalty, together with the curated search-balanced dataset, is sufficient to produce efficient on-demand search behavior that generalizes beyond the training distribution."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"MMSearch-R1 uses reinforcement learning to train multimodal models for on-demand multi-turn internet search with image and text tools, outperforming same-size RAG baselines and matching larger ones while cutting search calls by over 30%."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Reinforcement learning lets multimodal models search the internet only when needed"}],"snapshot_sha256":"3447c2a81aad5ee45d9eec2cce236f98202eda62d8441b689c80c1450340ee79"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"Robust deployment of large multimodal models (LMMs) in real-world scenarios requires access to external knowledge sources, given the complexity and dynamic nature of real-world information. Existing approaches such as retrieval-augmented generation (RAG) and prompt engineered search agents rely on rigid pipelines, often leading to inefficient or excessive search behaviors. We present MMSearch-R1, the first end-to-end reinforcement learning framework that enables LMMs to perform on-demand, multi-turn search in real-world Internet environments. Our framework integrates both image and text search","authors_text":"Bo Li, Bo You, Jinming Wu, Wei Li, Yiding Liu, Zejun Ma, Zihao Deng, Ziwei Liu","cross_cats":["cs.CL"],"headline":"Reinforcement learning lets multimodal models search the internet only when needed","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-06-25T17:59:42Z","title":"MMSearch-R1: Incentivizing LMMs to Search"},"references":{"count":94,"internal_anchors":26,"resolved_work":94,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Open deep search: Democratizing search with open-source reasoning agents","work_id":"467e5b5b-a169-4539-8ddc-bd34e07b0e3a","year":2025},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Claude 3.5 Sonnet","work_id":"45d58c07-b237-4bb8-ac8d-8c68f4784507","year":2024},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Self-rag: Learn- ing to retrieve, generate, and critique through self-reflection","work_id":"cd9884d5-53dc-4ba6-98d6-a38b02cf052e","year":2023},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Mint-1t: Scaling open-source multimodal data by 10x: A multimodal dataset with one trillion tokens","work_id":"9350a215-8bdd-4a04-a217-295413298f83","year":2024},{"cited_arxiv_id":"2502.13923","doi":"","is_internal_anchor":true,"ref_index":5,"title":"Qwen2.5-VL Technical Report","work_id":"69dffacb-bfe8-442d-be86-48624c60426f","year":2025}],"snapshot_sha256":"2a80b22b2e8323d5b48d357cc8e67cc82540c731f2a5ec48516d7e6cffe9fca1"},"source":{"id":"2506.20670","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-16T15:21:56.253302Z","id":"18f9ea2e-53a7-4c65-865b-759ce7bedc8c","model_set":{"reader":"grok-4.3"},"one_line_summary":"MMSearch-R1 uses reinforcement learning to train multimodal models for on-demand multi-turn internet search with image and text tools, outperforming same-size RAG baselines and matching larger ones while cutting search calls by over 30%.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Reinforcement learning lets multimodal models search the internet only when needed","strongest_claim":"We present MMSearch-R1, the first end-to-end reinforcement learning framework that enables LMMs to perform on-demand, multi-turn search in real-world Internet environments... our model not only outperforms RAG-based baselines of the same model size, but also matches the performance of a larger RAG-based model while reducing search calls by over 30%.","weakest_assumption":"The outcome-based reward combined with a search penalty, together with the curated search-balanced dataset, is sufficient to produce efficient on-demand search behavior that generalizes beyond the training distribution."}},"verdict_id":"18f9ea2e-53a7-4c65-865b-759ce7bedc8c"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:839be17c1a7c264f7a32fc413c22b3b36cc6138c3552fc6ecae4b151be147243","target":"record","created_at":"2026-05-17T23:38:47Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"d87506b60c8c6cbcb204c48af3d47e2d19d61c1bbb8a14d3a703a876367c7425","cross_cats_sorted":["cs.CL"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-06-25T17:59:42Z","title_canon_sha256":"8ecc7b053022d2d258e49fd9c9a94ad2866fd3f21cc2e0a15ea9923f28eb1f08"},"schema_version":"1.0","source":{"id":"2506.20670","kind":"arxiv","version":1}},"canonical_sha256":"669502c279c78d8099407fa0ff8f19a5a7a416364ed7208ee54d159fe652127e","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"669502c279c78d8099407fa0ff8f19a5a7a416364ed7208ee54d159fe652127e","first_computed_at":"2026-05-17T23:38:47.444503Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:47.444503Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"qekVEaEoh4rFM9FM5LvrDkpfPnAlMrKGR9suWfXA4LF4aQ4eHd1qSYhOp4/Blj0nIWEuc3i2OTcH3zKKL3u8CQ==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:47.444920Z","signed_message":"canonical_sha256_bytes"},"source_id":"2506.20670","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:839be17c1a7c264f7a32fc413c22b3b36cc6138c3552fc6ecae4b151be147243","sha256:f72e4d3b8c0be2bd3f472f32b8a82620a78675619a700dc04441c9a03635c536"],"state_sha256":"5a59e9383a0b60182cfbda9ebe2705023c6cabc3afd9e919a771bb62635aa7d8"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"HM74AmXnc8JUjm1mUz1Eu2swx/cQK6aNnCj/iAJxCE6PoPnaL0qpcZzQa7XN9XfmUH1Pjy5gs04+zYgwRwh8Bw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-05T09:44:16.157168Z","bundle_sha256":"efb53420f51dda9c69e810e4a6de20000dce6d9635d862b8b5a23732fe737105"}}