{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:RLLBD3WSTZA47IHNV3WVL7XON4","short_pith_number":"pith:RLLBD3WS","schema_version":"1.0","canonical_sha256":"8ad611eed29e41cfa0edaeed55feee6f247d2e50a053c620d1a2163523e6bc69","source":{"kind":"arxiv","id":"2507.21433","version":3},"attestation_state":"computed","paper":{"title":"ReasonCache: Accelerating Large Reasoning Model Serving through KV Cache Sharing","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Hong Xu, Jingzong Li, Kaiwen Chen, Minchen Yu, Xin Tan","submitted_at":"2025-07-29T02:05:51Z","abstract_excerpt":"Large Reasoning Models (LRMs) are becoming integral to many AI inference systems, enhancing their capabilities with advanced reasoning. However, deploying these models in production environments presents a significant QoS challenge: the substantial memory overhead from their long, auto-regressive inference processes severely limits throughput and increases latency, thereby affecting the quality of service for concurrent users. We observe that LRMs frequently generate highly similar intermediate reasoning steps, which, in turn, correspond to highly similar KV cache states across layers. Buildin"},"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":"2507.21433","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-07-29T02:05:51Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"1ee423a8466bd8d6169947ba044f624f2219043a49246dd5c9a5316c19f43637","abstract_canon_sha256":"de9bf8c21d8a10d265b7608feb46c5eb3ba5c796f6080cd5b4df8ecd10a9915e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:39:17.798668Z","signature_b64":"Vz3tB7/l4+H1TJ1N7wJO789GcB1tpyxEbxuq/bB4NWSij30amJIFooEWJszNxkXqqMJ19k5lInvZfC+P8Rk9BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8ad611eed29e41cfa0edaeed55feee6f247d2e50a053c620d1a2163523e6bc69","last_reissued_at":"2026-05-17T23:39:17.798003Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:39:17.798003Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"ReasonCache: Accelerating Large Reasoning Model Serving through KV Cache Sharing","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Hong Xu, Jingzong Li, Kaiwen Chen, Minchen Yu, Xin Tan","submitted_at":"2025-07-29T02:05:51Z","abstract_excerpt":"Large Reasoning Models (LRMs) are becoming integral to many AI inference systems, enhancing their capabilities with advanced reasoning. However, deploying these models in production environments presents a significant QoS challenge: the substantial memory overhead from their long, auto-regressive inference processes severely limits throughput and increases latency, thereby affecting the quality of service for concurrent users. We observe that LRMs frequently generate highly similar intermediate reasoning steps, which, in turn, correspond to highly similar KV cache states across layers. Buildin"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.21433","kind":"arxiv","version":3},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"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":"2507.21433","created_at":"2026-05-17T23:39:17.798092+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.21433v3","created_at":"2026-05-17T23:39:17.798092+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.21433","created_at":"2026-05-17T23:39:17.798092+00:00"},{"alias_kind":"pith_short_12","alias_value":"RLLBD3WSTZA4","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"RLLBD3WSTZA47IHN","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"RLLBD3WS","created_at":"2026-05-18T12:33:37.589309+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/RLLBD3WSTZA47IHNV3WVL7XON4","json":"https://pith.science/pith/RLLBD3WSTZA47IHNV3WVL7XON4.json","graph_json":"https://pith.science/api/pith-number/RLLBD3WSTZA47IHNV3WVL7XON4/graph.json","events_json":"https://pith.science/api/pith-number/RLLBD3WSTZA47IHNV3WVL7XON4/events.json","paper":"https://pith.science/paper/RLLBD3WS"},"agent_actions":{"view_html":"https://pith.science/pith/RLLBD3WSTZA47IHNV3WVL7XON4","download_json":"https://pith.science/pith/RLLBD3WSTZA47IHNV3WVL7XON4.json","view_paper":"https://pith.science/paper/RLLBD3WS","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.21433&json=true","fetch_graph":"https://pith.science/api/pith-number/RLLBD3WSTZA47IHNV3WVL7XON4/graph.json","fetch_events":"https://pith.science/api/pith-number/RLLBD3WSTZA47IHNV3WVL7XON4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RLLBD3WSTZA47IHNV3WVL7XON4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RLLBD3WSTZA47IHNV3WVL7XON4/action/storage_attestation","attest_author":"https://pith.science/pith/RLLBD3WSTZA47IHNV3WVL7XON4/action/author_attestation","sign_citation":"https://pith.science/pith/RLLBD3WSTZA47IHNV3WVL7XON4/action/citation_signature","submit_replication":"https://pith.science/pith/RLLBD3WSTZA47IHNV3WVL7XON4/action/replication_record"}},"created_at":"2026-05-17T23:39:17.798092+00:00","updated_at":"2026-05-17T23:39:17.798092+00:00"}