{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:UZUFVG7XVTPN62NB6C2MKNIWFO","short_pith_number":"pith:UZUFVG7X","schema_version":"1.0","canonical_sha256":"a6685a9bf7acdedf69a1f0b4c535162b8559f3ad17f4d0aa8f6e20331c143588","source":{"kind":"arxiv","id":"2607.19214","version":1},"attestation_state":"computed","paper":{"title":"Keeping the Cache Warm Pays: Keepalive Economics for Agentic Workloads","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Maxim Khailo","submitted_at":"2026-07-21T15:45:47Z","abstract_excerpt":"Frontier LLM providers cache a prompt's processed prefix so that a follow-up request sharing it pays ~10% of the input price and skips most of the prefill latency. Agentic workloads systematically destroy this benefit: the agent sends a request, runs a tool or waits for approval for minutes, and by the time the follow-up is sent the cached prefix has been evicted, so the agent pays the full prefill again. A client-side keepalive, replaying the prefix on a timer during the pause, prevents this, and it is individually rational: across Anthropic, OpenAI, Google, and DeepSeek we show that a keepal"},"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":"2607.19214","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.DC","submitted_at":"2026-07-21T15:45:47Z","cross_cats_sorted":[],"title_canon_sha256":"726bc9b25c49538ab9aacdce941af9be1ade77d203ea88728fb06c0bab3aab25","abstract_canon_sha256":"e2e4db7d704c984751b6af013cacb192a48400326b0da4ea688ffd6b5a329429"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-22T01:24:14.704666Z","signature_b64":"CAVVgqzQg+YESeqmz8szJxJCmEz6jdiam3UelLD/5ihTuV7pT32NKw4ln4k0g5cVnq5HjfPj9TxmO9qnli/yBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a6685a9bf7acdedf69a1f0b4c535162b8559f3ad17f4d0aa8f6e20331c143588","last_reissued_at":"2026-07-22T01:24:14.703909Z","signature_status":"signed_v1","first_computed_at":"2026-07-22T01:24:14.703909Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Keeping the Cache Warm Pays: Keepalive Economics for Agentic Workloads","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Maxim Khailo","submitted_at":"2026-07-21T15:45:47Z","abstract_excerpt":"Frontier LLM providers cache a prompt's processed prefix so that a follow-up request sharing it pays ~10% of the input price and skips most of the prefill latency. Agentic workloads systematically destroy this benefit: the agent sends a request, runs a tool or waits for approval for minutes, and by the time the follow-up is sent the cached prefix has been evicted, so the agent pays the full prefill again. A client-side keepalive, replaying the prefix on a timer during the pause, prevents this, and it is individually rational: across Anthropic, OpenAI, Google, and DeepSeek we show that a keepal"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.19214","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/2607.19214/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":"2607.19214","created_at":"2026-07-22T01:24:14.704321+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.19214v1","created_at":"2026-07-22T01:24:14.704321+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.19214","created_at":"2026-07-22T01:24:14.704321+00:00"},{"alias_kind":"pith_short_12","alias_value":"UZUFVG7XVTPN","created_at":"2026-07-22T01:24:14.704321+00:00"},{"alias_kind":"pith_short_16","alias_value":"UZUFVG7XVTPN62NB","created_at":"2026-07-22T01:24:14.704321+00:00"},{"alias_kind":"pith_short_8","alias_value":"UZUFVG7X","created_at":"2026-07-22T01:24:14.704321+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/UZUFVG7XVTPN62NB6C2MKNIWFO","json":"https://pith.science/pith/UZUFVG7XVTPN62NB6C2MKNIWFO.json","graph_json":"https://pith.science/api/pith-number/UZUFVG7XVTPN62NB6C2MKNIWFO/graph.json","events_json":"https://pith.science/api/pith-number/UZUFVG7XVTPN62NB6C2MKNIWFO/events.json","paper":"https://pith.science/paper/UZUFVG7X"},"agent_actions":{"view_html":"https://pith.science/pith/UZUFVG7XVTPN62NB6C2MKNIWFO","download_json":"https://pith.science/pith/UZUFVG7XVTPN62NB6C2MKNIWFO.json","view_paper":"https://pith.science/paper/UZUFVG7X","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.19214&json=true","fetch_graph":"https://pith.science/api/pith-number/UZUFVG7XVTPN62NB6C2MKNIWFO/graph.json","fetch_events":"https://pith.science/api/pith-number/UZUFVG7XVTPN62NB6C2MKNIWFO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UZUFVG7XVTPN62NB6C2MKNIWFO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UZUFVG7XVTPN62NB6C2MKNIWFO/action/storage_attestation","attest_author":"https://pith.science/pith/UZUFVG7XVTPN62NB6C2MKNIWFO/action/author_attestation","sign_citation":"https://pith.science/pith/UZUFVG7XVTPN62NB6C2MKNIWFO/action/citation_signature","submit_replication":"https://pith.science/pith/UZUFVG7XVTPN62NB6C2MKNIWFO/action/replication_record"}},"created_at":"2026-07-22T01:24:14.704321+00:00","updated_at":"2026-07-22T01:24:14.704321+00:00"}