{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:OOYBLIWOC7Z4BVHIRHWESVUJJ3","short_pith_number":"pith:OOYBLIWO","schema_version":"1.0","canonical_sha256":"73b015a2ce17f3c0d4e889ec4956894ed45c44c9fdd4ae3f407d9e1c94d26f3a","source":{"kind":"arxiv","id":"2606.26836","version":1},"attestation_state":"computed","paper":{"title":"The Capability Frontier: Benchmarks Miss 82% of Model Performance","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Amirali Abdullah, Ant\\'ia Garc\\'ia, Bradley Fowler, Daniel Thi Graviet, Fazl Barez, Joshua Greaves, Narmeen Fatimah Oozeer, Philip Quirke, Ryan Smith, Shriyash Kaustubh Upadhyay, William Myers","submitted_at":"2026-06-25T10:20:47Z","abstract_excerpt":"Existing benchmarks typically report accuracy for a single model on a single run. This systematically understates real-world LLM capabilities, particularly under heterogeneous data distributions: (i) different models get different questions correct according to their specializations, and (ii) given a budget, multiple generations can be sampled and selectively retained. To quantify this gap, we introduce the Capability Frontier: a Pareto frontier over a set of models that characterizes the best achievable performance at each cost level under optimal selection across models and generations (i.e."},"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":"2606.26836","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-06-25T10:20:47Z","cross_cats_sorted":[],"title_canon_sha256":"3fced67521c802376d8b52472991cb04a0c79cf04ebd460565e5321cb31f47c5","abstract_canon_sha256":"cab566424273db9fc094f4ff013daa214ccd3184ce33999b13c4aa1127dbcd8e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-26T01:16:01.250284Z","signature_b64":"d6YvC2q+qCrPjMV/N82I2hEFsMV54iotMIeWFVU+n94HThJj/G2O96LWijHNaFtIAvFpVtDPd5i5GJYRd6gDBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"73b015a2ce17f3c0d4e889ec4956894ed45c44c9fdd4ae3f407d9e1c94d26f3a","last_reissued_at":"2026-06-26T01:16:01.249876Z","signature_status":"signed_v1","first_computed_at":"2026-06-26T01:16:01.249876Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"The Capability Frontier: Benchmarks Miss 82% of Model Performance","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Amirali Abdullah, Ant\\'ia Garc\\'ia, Bradley Fowler, Daniel Thi Graviet, Fazl Barez, Joshua Greaves, Narmeen Fatimah Oozeer, Philip Quirke, Ryan Smith, Shriyash Kaustubh Upadhyay, William Myers","submitted_at":"2026-06-25T10:20:47Z","abstract_excerpt":"Existing benchmarks typically report accuracy for a single model on a single run. This systematically understates real-world LLM capabilities, particularly under heterogeneous data distributions: (i) different models get different questions correct according to their specializations, and (ii) given a budget, multiple generations can be sampled and selectively retained. To quantify this gap, we introduce the Capability Frontier: a Pareto frontier over a set of models that characterizes the best achievable performance at each cost level under optimal selection across models and generations (i.e."},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.26836","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/2606.26836/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":"2606.26836","created_at":"2026-06-26T01:16:01.249933+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.26836v1","created_at":"2026-06-26T01:16:01.249933+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.26836","created_at":"2026-06-26T01:16:01.249933+00:00"},{"alias_kind":"pith_short_12","alias_value":"OOYBLIWOC7Z4","created_at":"2026-06-26T01:16:01.249933+00:00"},{"alias_kind":"pith_short_16","alias_value":"OOYBLIWOC7Z4BVHI","created_at":"2026-06-26T01:16:01.249933+00:00"},{"alias_kind":"pith_short_8","alias_value":"OOYBLIWO","created_at":"2026-06-26T01:16:01.249933+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/OOYBLIWOC7Z4BVHIRHWESVUJJ3","json":"https://pith.science/pith/OOYBLIWOC7Z4BVHIRHWESVUJJ3.json","graph_json":"https://pith.science/api/pith-number/OOYBLIWOC7Z4BVHIRHWESVUJJ3/graph.json","events_json":"https://pith.science/api/pith-number/OOYBLIWOC7Z4BVHIRHWESVUJJ3/events.json","paper":"https://pith.science/paper/OOYBLIWO"},"agent_actions":{"view_html":"https://pith.science/pith/OOYBLIWOC7Z4BVHIRHWESVUJJ3","download_json":"https://pith.science/pith/OOYBLIWOC7Z4BVHIRHWESVUJJ3.json","view_paper":"https://pith.science/paper/OOYBLIWO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.26836&json=true","fetch_graph":"https://pith.science/api/pith-number/OOYBLIWOC7Z4BVHIRHWESVUJJ3/graph.json","fetch_events":"https://pith.science/api/pith-number/OOYBLIWOC7Z4BVHIRHWESVUJJ3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OOYBLIWOC7Z4BVHIRHWESVUJJ3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OOYBLIWOC7Z4BVHIRHWESVUJJ3/action/storage_attestation","attest_author":"https://pith.science/pith/OOYBLIWOC7Z4BVHIRHWESVUJJ3/action/author_attestation","sign_citation":"https://pith.science/pith/OOYBLIWOC7Z4BVHIRHWESVUJJ3/action/citation_signature","submit_replication":"https://pith.science/pith/OOYBLIWOC7Z4BVHIRHWESVUJJ3/action/replication_record"}},"created_at":"2026-06-26T01:16:01.249933+00:00","updated_at":"2026-06-26T01:16:01.249933+00:00"}