{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:KLPNNOSVBBHPZTR3YKEPHY5ZHX","short_pith_number":"pith:KLPNNOSV","schema_version":"1.0","canonical_sha256":"52ded6ba55084efcce3bc288f3e3b93dc86032106d4a68852d1863caf969f5e1","source":{"kind":"arxiv","id":"2605.19099","version":1},"attestation_state":"computed","paper":{"title":"DecisionBench: A Benchmark for Emergent Delegation in Long-Horizon Agentic Workflows","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.MA"],"primary_cat":"cs.AI","authors_text":"Ao Qu, Megan Wang, Yi Ling Yu, Yuxuan Gao, Zijian Carl Ma","submitted_at":"2026-05-18T20:37:14Z","abstract_excerpt":"We introduce DecisionBench, a benchmark substrate for emergent delegation in long-horizon agentic workflows. The substrate fixes a task suite (GAIA, tau-bench, BFCL multi-turn), a peer-model pool (11 models, 7 vendor families), a delegation interface (call_model plus an optional read_profile channel), a deterministic skill-annotation layer, and a multi-axis metric suite covering quality, cost, latency, delegation rate, routing fidelity-at-k, vendor self-preference, and a counterfactual-delegation ceiling. The substrate is agnostic to how peer information is generated or delivered, so learned r"},"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":"2605.19099","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-18T20:37:14Z","cross_cats_sorted":["cs.CL","cs.MA"],"title_canon_sha256":"dfef9b72fafb337dfa8aa35eea71a39706d1018de50152615c4ca7a8aec20be6","abstract_canon_sha256":"73a886871672c8fcfa9e6b7f5318e7e54bca6f9b9c424641f9eda802b033af8d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T01:05:27.106255Z","signature_b64":"k0L/+4U2l3/O4dhNKiZz3cPsBOGzMFE/4IiRhzNQdObvEvnW3SAWhe+QlL8G+71MlXLlYP2IvsCbIbNFXf3hDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"52ded6ba55084efcce3bc288f3e3b93dc86032106d4a68852d1863caf969f5e1","last_reissued_at":"2026-05-20T01:05:27.105521Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T01:05:27.105521Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DecisionBench: A Benchmark for Emergent Delegation in Long-Horizon Agentic Workflows","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.MA"],"primary_cat":"cs.AI","authors_text":"Ao Qu, Megan Wang, Yi Ling Yu, Yuxuan Gao, Zijian Carl Ma","submitted_at":"2026-05-18T20:37:14Z","abstract_excerpt":"We introduce DecisionBench, a benchmark substrate for emergent delegation in long-horizon agentic workflows. The substrate fixes a task suite (GAIA, tau-bench, BFCL multi-turn), a peer-model pool (11 models, 7 vendor families), a delegation interface (call_model plus an optional read_profile channel), a deterministic skill-annotation layer, and a multi-axis metric suite covering quality, cost, latency, delegation rate, routing fidelity-at-k, vendor self-preference, and a counterfactual-delegation ceiling. The substrate is agnostic to how peer information is generated or delivered, so learned r"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.19099","kind":"arxiv","version":1},"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/2605.19099/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":"2605.19099","created_at":"2026-05-20T01:05:27.105644+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.19099v1","created_at":"2026-05-20T01:05:27.105644+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.19099","created_at":"2026-05-20T01:05:27.105644+00:00"},{"alias_kind":"pith_short_12","alias_value":"KLPNNOSVBBHP","created_at":"2026-05-20T01:05:27.105644+00:00"},{"alias_kind":"pith_short_16","alias_value":"KLPNNOSVBBHPZTR3","created_at":"2026-05-20T01:05:27.105644+00:00"},{"alias_kind":"pith_short_8","alias_value":"KLPNNOSV","created_at":"2026-05-20T01:05:27.105644+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KLPNNOSVBBHPZTR3YKEPHY5ZHX","json":"https://pith.science/pith/KLPNNOSVBBHPZTR3YKEPHY5ZHX.json","graph_json":"https://pith.science/api/pith-number/KLPNNOSVBBHPZTR3YKEPHY5ZHX/graph.json","events_json":"https://pith.science/api/pith-number/KLPNNOSVBBHPZTR3YKEPHY5ZHX/events.json","paper":"https://pith.science/paper/KLPNNOSV"},"agent_actions":{"view_html":"https://pith.science/pith/KLPNNOSVBBHPZTR3YKEPHY5ZHX","download_json":"https://pith.science/pith/KLPNNOSVBBHPZTR3YKEPHY5ZHX.json","view_paper":"https://pith.science/paper/KLPNNOSV","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.19099&json=true","fetch_graph":"https://pith.science/api/pith-number/KLPNNOSVBBHPZTR3YKEPHY5ZHX/graph.json","fetch_events":"https://pith.science/api/pith-number/KLPNNOSVBBHPZTR3YKEPHY5ZHX/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KLPNNOSVBBHPZTR3YKEPHY5ZHX/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KLPNNOSVBBHPZTR3YKEPHY5ZHX/action/storage_attestation","attest_author":"https://pith.science/pith/KLPNNOSVBBHPZTR3YKEPHY5ZHX/action/author_attestation","sign_citation":"https://pith.science/pith/KLPNNOSVBBHPZTR3YKEPHY5ZHX/action/citation_signature","submit_replication":"https://pith.science/pith/KLPNNOSVBBHPZTR3YKEPHY5ZHX/action/replication_record"}},"created_at":"2026-05-20T01:05:27.105644+00:00","updated_at":"2026-05-20T01:05:27.105644+00:00"}