{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:WXGAP5MECDC5CAS6P4EPXYLTHL","short_pith_number":"pith:WXGAP5ME","schema_version":"1.0","canonical_sha256":"b5cc07f58410c5d1025e7f08fbe1733ae7b3173c7e006e7d79f555507630bb40","source":{"kind":"arxiv","id":"2502.15964","version":1},"attestation_state":"computed","paper":{"title":"Minions: Cost-efficient Collaboration Between On-device and Cloud Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.DC"],"primary_cat":"cs.LG","authors_text":"Avanika Narayan, Avner May, Christopher Re, Dan Biderman, James Zou, Sabri Eyuboglu, Scott Linderman","submitted_at":"2025-02-21T21:54:40Z","abstract_excerpt":"We investigate an emerging setup in which a small, on-device language model (LM) with access to local data communicates with a frontier, cloud-hosted LM to solve real-world tasks involving financial, medical, and scientific reasoning over long documents. Can a local-remote collaboration reduce cloud inference costs while preserving quality? First, we consider a naive collaboration protocol where the local and remote models simply chat back and forth. Because only the local model reads the full context, this protocol achieves a 30.4x reduction in remote costs, but recovers only 87% of the perfo"},"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":"2502.15964","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-02-21T21:54:40Z","cross_cats_sorted":["cs.AI","cs.CL","cs.DC"],"title_canon_sha256":"80314a219f01aeba9de43bfba6eddc7e1b9afe5bd3b1fd28b1fc6a6759d57185","abstract_canon_sha256":"62e79f313f2af8dda4780c48f46b9c40bed16e7b92acf3500c5ced4ed3523489"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:20:19.907481Z","signature_b64":"vqjXYonJQ0fy7trhdRk6hQsZMGKaDzqxvm6hzvt9PyvwgZNefs6At15DRipynJcWG5WAbmgictXsbeeaTx4JBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b5cc07f58410c5d1025e7f08fbe1733ae7b3173c7e006e7d79f555507630bb40","last_reissued_at":"2026-07-05T10:20:19.906972Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:20:19.906972Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Minions: Cost-efficient Collaboration Between On-device and Cloud Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.DC"],"primary_cat":"cs.LG","authors_text":"Avanika Narayan, Avner May, Christopher Re, Dan Biderman, James Zou, Sabri Eyuboglu, Scott Linderman","submitted_at":"2025-02-21T21:54:40Z","abstract_excerpt":"We investigate an emerging setup in which a small, on-device language model (LM) with access to local data communicates with a frontier, cloud-hosted LM to solve real-world tasks involving financial, medical, and scientific reasoning over long documents. Can a local-remote collaboration reduce cloud inference costs while preserving quality? First, we consider a naive collaboration protocol where the local and remote models simply chat back and forth. Because only the local model reads the full context, this protocol achieves a 30.4x reduction in remote costs, but recovers only 87% of the perfo"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.15964","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/2502.15964/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":"2502.15964","created_at":"2026-07-05T10:20:19.907030+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.15964v1","created_at":"2026-07-05T10:20:19.907030+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.15964","created_at":"2026-07-05T10:20:19.907030+00:00"},{"alias_kind":"pith_short_12","alias_value":"WXGAP5MECDC5","created_at":"2026-07-05T10:20:19.907030+00:00"},{"alias_kind":"pith_short_16","alias_value":"WXGAP5MECDC5CAS6","created_at":"2026-07-05T10:20:19.907030+00:00"},{"alias_kind":"pith_short_8","alias_value":"WXGAP5ME","created_at":"2026-07-05T10:20:19.907030+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.13643","citing_title":"Recursive Agent Harnesses","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20630","citing_title":"Evaluating Temporal Semantic Caching and Workflow Optimization in Agentic Plan-Execute Pipelines","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17172","citing_title":"OpenJarvis: Personal AI, On Personal Devices","ref_index":60,"is_internal_anchor":false},{"citing_arxiv_id":"2511.22599","citing_title":"DisCEdge: Distributed Context Management for Large Language Models at the Edge","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12946","citing_title":"Parcae: Scaling Laws For Stable Looped Language Models","ref_index":56,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/WXGAP5MECDC5CAS6P4EPXYLTHL","json":"https://pith.science/pith/WXGAP5MECDC5CAS6P4EPXYLTHL.json","graph_json":"https://pith.science/api/pith-number/WXGAP5MECDC5CAS6P4EPXYLTHL/graph.json","events_json":"https://pith.science/api/pith-number/WXGAP5MECDC5CAS6P4EPXYLTHL/events.json","paper":"https://pith.science/paper/WXGAP5ME"},"agent_actions":{"view_html":"https://pith.science/pith/WXGAP5MECDC5CAS6P4EPXYLTHL","download_json":"https://pith.science/pith/WXGAP5MECDC5CAS6P4EPXYLTHL.json","view_paper":"https://pith.science/paper/WXGAP5ME","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.15964&json=true","fetch_graph":"https://pith.science/api/pith-number/WXGAP5MECDC5CAS6P4EPXYLTHL/graph.json","fetch_events":"https://pith.science/api/pith-number/WXGAP5MECDC5CAS6P4EPXYLTHL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WXGAP5MECDC5CAS6P4EPXYLTHL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WXGAP5MECDC5CAS6P4EPXYLTHL/action/storage_attestation","attest_author":"https://pith.science/pith/WXGAP5MECDC5CAS6P4EPXYLTHL/action/author_attestation","sign_citation":"https://pith.science/pith/WXGAP5MECDC5CAS6P4EPXYLTHL/action/citation_signature","submit_replication":"https://pith.science/pith/WXGAP5MECDC5CAS6P4EPXYLTHL/action/replication_record"}},"created_at":"2026-07-05T10:20:19.907030+00:00","updated_at":"2026-07-05T10:20:19.907030+00:00"}