{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:MB4L66YLHIUOAJABVQGPH75TMZ","short_pith_number":"pith:MB4L66YL","schema_version":"1.0","canonical_sha256":"6078bf7b0b3a28e02401ac0cf3ffb366711a029e929681449ef097bc90219513","source":{"kind":"arxiv","id":"2605.20315","version":1},"attestation_state":"computed","paper":{"title":"Mix-Quant: Quantized Prefilling, Precise Decoding for Agentic LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Gongfan Fang, Haiquan Lu, Xinchao Wang, Xinyin Ma, Zigeng Chen","submitted_at":"2026-05-19T17:50:17Z","abstract_excerpt":"LLM agents have recently emerged as a powerful paradigm for solving complex tasks through planning, tool use, memory retrieval, and multi-step interaction. However, these agentic workflows often introduce substantial input-side overhead, making the compute-intensive prefilling stage a key bottleneck in long-context, multi-turn inference. In this work, we propose Mix-Quant, a simple and effective phase-aware quantization framework for fast agentic inference. We first investigate FP4 quantization in agentic LLM workflows and observe that quantizing the entire inference process can incur signific"},"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":"2605.20315","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-05-19T17:50:17Z","cross_cats_sorted":[],"title_canon_sha256":"fb5d56ae28b70bb6f778782c2ef52aa8128bfb06c1cf7c53b3f4533a3d2e24dc","abstract_canon_sha256":"17660a1b435a431cd0e34b78e99a56be11670fae723d9675624c72582ce6968b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-21T00:04:25.070565Z","signature_b64":"RMCYmvj7WHUlPVMvwzqgULdnX5EIHj97SmjyzoxzES7YtEhGWwmwZLSV4s75a1L/m6/RPfSElAXGv2dB68PoCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6078bf7b0b3a28e02401ac0cf3ffb366711a029e929681449ef097bc90219513","last_reissued_at":"2026-05-21T00:04:25.069721Z","signature_status":"signed_v1","first_computed_at":"2026-05-21T00:04:25.069721Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Mix-Quant: Quantized Prefilling, Precise Decoding for Agentic LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Gongfan Fang, Haiquan Lu, Xinchao Wang, Xinyin Ma, Zigeng Chen","submitted_at":"2026-05-19T17:50:17Z","abstract_excerpt":"LLM agents have recently emerged as a powerful paradigm for solving complex tasks through planning, tool use, memory retrieval, and multi-step interaction. However, these agentic workflows often introduce substantial input-side overhead, making the compute-intensive prefilling stage a key bottleneck in long-context, multi-turn inference. In this work, we propose Mix-Quant, a simple and effective phase-aware quantization framework for fast agentic inference. We first investigate FP4 quantization in agentic LLM workflows and observe that quantizing the entire inference process can incur signific"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.20315","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/2605.20315/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":"2605.20315","created_at":"2026-05-21T00:04:25.069823+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.20315v1","created_at":"2026-05-21T00:04:25.069823+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.20315","created_at":"2026-05-21T00:04:25.069823+00:00"},{"alias_kind":"pith_short_12","alias_value":"MB4L66YLHIUO","created_at":"2026-05-21T00:04:25.069823+00:00"},{"alias_kind":"pith_short_16","alias_value":"MB4L66YLHIUOAJAB","created_at":"2026-05-21T00:04:25.069823+00:00"},{"alias_kind":"pith_short_8","alias_value":"MB4L66YL","created_at":"2026-05-21T00:04:25.069823+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/MB4L66YLHIUOAJABVQGPH75TMZ","json":"https://pith.science/pith/MB4L66YLHIUOAJABVQGPH75TMZ.json","graph_json":"https://pith.science/api/pith-number/MB4L66YLHIUOAJABVQGPH75TMZ/graph.json","events_json":"https://pith.science/api/pith-number/MB4L66YLHIUOAJABVQGPH75TMZ/events.json","paper":"https://pith.science/paper/MB4L66YL"},"agent_actions":{"view_html":"https://pith.science/pith/MB4L66YLHIUOAJABVQGPH75TMZ","download_json":"https://pith.science/pith/MB4L66YLHIUOAJABVQGPH75TMZ.json","view_paper":"https://pith.science/paper/MB4L66YL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.20315&json=true","fetch_graph":"https://pith.science/api/pith-number/MB4L66YLHIUOAJABVQGPH75TMZ/graph.json","fetch_events":"https://pith.science/api/pith-number/MB4L66YLHIUOAJABVQGPH75TMZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MB4L66YLHIUOAJABVQGPH75TMZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MB4L66YLHIUOAJABVQGPH75TMZ/action/storage_attestation","attest_author":"https://pith.science/pith/MB4L66YLHIUOAJABVQGPH75TMZ/action/author_attestation","sign_citation":"https://pith.science/pith/MB4L66YLHIUOAJABVQGPH75TMZ/action/citation_signature","submit_replication":"https://pith.science/pith/MB4L66YLHIUOAJABVQGPH75TMZ/action/replication_record"}},"created_at":"2026-05-21T00:04:25.069823+00:00","updated_at":"2026-05-21T00:04:25.069823+00:00"}