{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:5B5OBF7WEU652PIUBN26WBE5SX","short_pith_number":"pith:5B5OBF7W","canonical_record":{"source":{"id":"2504.12516","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-04-16T22:27:45Z","cross_cats_sorted":[],"title_canon_sha256":"21b2eafc19575259cd304f98bd064fd45e66c59e048165d0ce41f3aeb023acbc","abstract_canon_sha256":"c69d837c58a9126cb1e741272931c069fa1ea9a387b1d7d56c7610b540ab72e0"},"schema_version":"1.0"},"canonical_sha256":"e87ae097f6253ddd3d140b75eb049d95dd965a6057de563d67f98ac73847888d","source":{"kind":"arxiv","id":"2504.12516","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2504.12516","created_at":"2026-07-05T10:50:17Z"},{"alias_kind":"arxiv_version","alias_value":"2504.12516v1","created_at":"2026-07-05T10:50:17Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.12516","created_at":"2026-07-05T10:50:17Z"},{"alias_kind":"pith_short_12","alias_value":"5B5OBF7WEU65","created_at":"2026-07-05T10:50:17Z"},{"alias_kind":"pith_short_16","alias_value":"5B5OBF7WEU652PIU","created_at":"2026-07-05T10:50:17Z"},{"alias_kind":"pith_short_8","alias_value":"5B5OBF7W","created_at":"2026-07-05T10:50:17Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:5B5OBF7WEU652PIUBN26WBE5SX","target":"record","payload":{"canonical_record":{"source":{"id":"2504.12516","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-04-16T22:27:45Z","cross_cats_sorted":[],"title_canon_sha256":"21b2eafc19575259cd304f98bd064fd45e66c59e048165d0ce41f3aeb023acbc","abstract_canon_sha256":"c69d837c58a9126cb1e741272931c069fa1ea9a387b1d7d56c7610b540ab72e0"},"schema_version":"1.0"},"canonical_sha256":"e87ae097f6253ddd3d140b75eb049d95dd965a6057de563d67f98ac73847888d","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:50:17.710513Z","signature_b64":"7Tm1KZsV4JsAJixRmnsWs06/MILZZcpWbstQA5SOA0DO7acaEd4bVcKk+XPGUBivJsc6kkozDsNMiLwd5+8iBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e87ae097f6253ddd3d140b75eb049d95dd965a6057de563d67f98ac73847888d","last_reissued_at":"2026-07-05T10:50:17.709942Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:50:17.709942Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2504.12516","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":"2026-07-05T10:50:17Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"HCkBxDV/xY3HDhhUTH+cm9gE2jNqJmjd0/Zd5A9pgpb9k3+bVZLinGkLMjuiIVi26LWf3CsSRuWlVkfQkxVaAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-24T03:00:48.708162Z"},"content_sha256":"065155259d2c658f100af76f93284f93ea76a185ec3499951681bfffcbb2706f","schema_version":"1.0","event_id":"sha256:065155259d2c658f100af76f93284f93ea76a185ec3499951681bfffcbb2706f"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:5B5OBF7WEU652PIUBN26WBE5SX","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agents","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"BrowseComp offers 1,266 short-answer questions to test agents' persistence and creativity while browsing the web for hard-to-find information.","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Alex Tachard Passos, Amelia Glaese, Hyung Won Chung, Isa Fulford, Jason Wei, Jeffrey Han, Scott McKinney, Spencer Papay, William Fedus, Zhiqing Sun","submitted_at":"2025-04-16T22:27:45Z","abstract_excerpt":"We present BrowseComp, a simple yet challenging benchmark for measuring the ability for agents to browse the web. BrowseComp comprises 1,266 questions that require persistently navigating the internet in search of hard-to-find, entangled information. Despite the difficulty of the questions, BrowseComp is simple and easy-to-use, as predicted answers are short and easily verifiable against reference answers. BrowseComp for browsing agents can be seen as analogous to how programming competitions are an incomplete but useful benchmark for coding agents. While BrowseComp sidesteps challenges of a t"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"BrowseComp comprises 1,266 questions that require persistently navigating the internet in search of hard-to-find, entangled information and measures the important core capability of exercising persistence and creativity in finding information.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That short, easily verifiable answers are sufficient to measure the targeted capabilities of persistence and creativity, and that the chosen questions adequately represent the core challenges of real-world browsing without needing to address long-form answers or ambiguity.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"BrowseComp is a benchmark with 1,266 verifiable questions that test agents on persistent web browsing for entangled information.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"BrowseComp offers 1,266 short-answer questions to test agents' persistence and creativity while browsing the web for hard-to-find information.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"72047c0fa6855395c61c221821fd5ddffa7d0b06979e5cf8f8f94dbccb720428"},"source":{"id":"2504.12516","kind":"arxiv","version":1},"verdict":{"id":"121bbd6a-03c8-420f-878e-694d93941e66","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-12T07:39:42.916354Z","strongest_claim":"BrowseComp comprises 1,266 questions that require persistently navigating the internet in search of hard-to-find, entangled information and measures the important core capability of exercising persistence and creativity in finding information.","one_line_summary":"BrowseComp is a benchmark with 1,266 verifiable questions that test agents on persistent web browsing for entangled information.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That short, easily verifiable answers are sufficient to measure the targeted capabilities of persistence and creativity, and that the chosen questions adequately represent the core challenges of real-world browsing without needing to address long-form answers or ambiguity.","pith_extraction_headline":"BrowseComp offers 1,266 short-answer questions to test agents' persistence and creativity while browsing the web for hard-to-find information."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2504.12516/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":3,"snapshot_sha256":"9eea3e5fcbffca8a20b9e3ff246cb689465a4d981c75765f8d31bb82fd6880e7"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"121bbd6a-03c8-420f-878e-694d93941e66"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:50:17Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"VcaOw0af4h4dr4VIsx+Kcx33vy1f256RNZs6515BcYzhaC5CKlfyDISgnP1S8r8Kd2JLOrt939Bp9BE9LZh+Bg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-24T03:00:48.708937Z"},"content_sha256":"66731d438117c685379ad7766bb570afb06c33ca04e85d7dec090438c4800ac3","schema_version":"1.0","event_id":"sha256:66731d438117c685379ad7766bb570afb06c33ca04e85d7dec090438c4800ac3"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/5B5OBF7WEU652PIUBN26WBE5SX/bundle.json","state_url":"https://pith.science/pith/5B5OBF7WEU652PIUBN26WBE5SX/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/5B5OBF7WEU652PIUBN26WBE5SX/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":"2026-08-24T03:00:48Z","links":{"resolver":"https://pith.science/pith/5B5OBF7WEU652PIUBN26WBE5SX","bundle":"https://pith.science/pith/5B5OBF7WEU652PIUBN26WBE5SX/bundle.json","state":"https://pith.science/pith/5B5OBF7WEU652PIUBN26WBE5SX/state.json","well_known_bundle":"https://pith.science/.well-known/pith/5B5OBF7WEU652PIUBN26WBE5SX/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:5B5OBF7WEU652PIUBN26WBE5SX","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":"c69d837c58a9126cb1e741272931c069fa1ea9a387b1d7d56c7610b540ab72e0","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-04-16T22:27:45Z","title_canon_sha256":"21b2eafc19575259cd304f98bd064fd45e66c59e048165d0ce41f3aeb023acbc"},"schema_version":"1.0","source":{"id":"2504.12516","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2504.12516","created_at":"2026-07-05T10:50:17Z"},{"alias_kind":"arxiv_version","alias_value":"2504.12516v1","created_at":"2026-07-05T10:50:17Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.12516","created_at":"2026-07-05T10:50:17Z"},{"alias_kind":"pith_short_12","alias_value":"5B5OBF7WEU65","created_at":"2026-07-05T10:50:17Z"},{"alias_kind":"pith_short_16","alias_value":"5B5OBF7WEU652PIU","created_at":"2026-07-05T10:50:17Z"},{"alias_kind":"pith_short_8","alias_value":"5B5OBF7W","created_at":"2026-07-05T10:50:17Z"}],"graph_snapshots":[{"event_id":"sha256:66731d438117c685379ad7766bb570afb06c33ca04e85d7dec090438c4800ac3","target":"graph","created_at":"2026-07-05T10:50:17Z","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":"BrowseComp comprises 1,266 questions that require persistently navigating the internet in search of hard-to-find, entangled information and measures the important core capability of exercising persistence and creativity in finding information."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That short, easily verifiable answers are sufficient to measure the targeted capabilities of persistence and creativity, and that the chosen questions adequately represent the core challenges of real-world browsing without needing to address long-form answers or ambiguity."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"BrowseComp is a benchmark with 1,266 verifiable questions that test agents on persistent web browsing for entangled information."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"BrowseComp offers 1,266 short-answer questions to test agents' persistence and creativity while browsing the web for hard-to-find information."}],"snapshot_sha256":"72047c0fa6855395c61c221821fd5ddffa7d0b06979e5cf8f8f94dbccb720428"},"formal_canon":{"evidence_count":3,"snapshot_sha256":"9eea3e5fcbffca8a20b9e3ff246cb689465a4d981c75765f8d31bb82fd6880e7"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2504.12516/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"We present BrowseComp, a simple yet challenging benchmark for measuring the ability for agents to browse the web. BrowseComp comprises 1,266 questions that require persistently navigating the internet in search of hard-to-find, entangled information. Despite the difficulty of the questions, BrowseComp is simple and easy-to-use, as predicted answers are short and easily verifiable against reference answers. BrowseComp for browsing agents can be seen as analogous to how programming competitions are an incomplete but useful benchmark for coding agents. While BrowseComp sidesteps challenges of a t","authors_text":"Alex Tachard Passos, Amelia Glaese, Hyung Won Chung, Isa Fulford, Jason Wei, Jeffrey Han, Scott McKinney, Spencer Papay, William Fedus, Zhiqing Sun","cross_cats":[],"headline":"BrowseComp offers 1,266 short-answer questions to test agents' persistence and creativity while browsing the web for hard-to-find information.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-04-16T22:27:45Z","title":"BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agents"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.12516","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-12T07:39:42.916354Z","id":"121bbd6a-03c8-420f-878e-694d93941e66","model_set":{"reader":"grok-4.3"},"one_line_summary":"BrowseComp is a benchmark with 1,266 verifiable questions that test agents on persistent web browsing for entangled information.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"BrowseComp offers 1,266 short-answer questions to test agents' persistence and creativity while browsing the web for hard-to-find information.","strongest_claim":"BrowseComp comprises 1,266 questions that require persistently navigating the internet in search of hard-to-find, entangled information and measures the important core capability of exercising persistence and creativity in finding information.","weakest_assumption":"That short, easily verifiable answers are sufficient to measure the targeted capabilities of persistence and creativity, and that the chosen questions adequately represent the core challenges of real-world browsing without needing to address long-form answers or ambiguity."}},"verdict_id":"121bbd6a-03c8-420f-878e-694d93941e66"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:065155259d2c658f100af76f93284f93ea76a185ec3499951681bfffcbb2706f","target":"record","created_at":"2026-07-05T10:50:17Z","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":"c69d837c58a9126cb1e741272931c069fa1ea9a387b1d7d56c7610b540ab72e0","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-04-16T22:27:45Z","title_canon_sha256":"21b2eafc19575259cd304f98bd064fd45e66c59e048165d0ce41f3aeb023acbc"},"schema_version":"1.0","source":{"id":"2504.12516","kind":"arxiv","version":1}},"canonical_sha256":"e87ae097f6253ddd3d140b75eb049d95dd965a6057de563d67f98ac73847888d","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"e87ae097f6253ddd3d140b75eb049d95dd965a6057de563d67f98ac73847888d","first_computed_at":"2026-07-05T10:50:17.709942Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:50:17.709942Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"7Tm1KZsV4JsAJixRmnsWs06/MILZZcpWbstQA5SOA0DO7acaEd4bVcKk+XPGUBivJsc6kkozDsNMiLwd5+8iBw==","signature_status":"signed_v1","signed_at":"2026-07-05T10:50:17.710513Z","signed_message":"canonical_sha256_bytes"},"source_id":"2504.12516","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:065155259d2c658f100af76f93284f93ea76a185ec3499951681bfffcbb2706f","sha256:66731d438117c685379ad7766bb570afb06c33ca04e85d7dec090438c4800ac3"],"state_sha256":"f49c33adfc39ef015b6793eb27f1b9a5730380506d30555ce31219c4537b25b3"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"+tHdUY96C/Tm2DCGf77tLfjABMqcwvqkeFK25cSaWoZ0e3WZC3e6soY09l4lq8V3OYJdETwURwdhsgjdOCtXBg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-24T03:00:48.715990Z","bundle_sha256":"b791953da049ef0dfa2f9427bd320c8f1578a244c92bcbc65111f7b3048e948d"}}