{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:AJOP2LMO45JELL324FTHE7XWUD","short_pith_number":"pith:AJOP2LMO","schema_version":"1.0","canonical_sha256":"025cfd2d8ee75245af7ae166727ef6a0ff5287b2feb8be4d271e69c004e7c266","source":{"kind":"arxiv","id":"2605.16480","version":1},"attestation_state":"computed","paper":{"title":"MoleCode unlocks structural intelligence in large language models","license":"http://creativecommons.org/publicdomain/zero/1.0/","headline":"MoleCode makes molecular topology directly readable, editable and auditable by LLMs instead of hidden in SMILES strings.","cross_cats":["cs.AI"],"primary_cat":"q-bio.BM","authors_text":"Boxuan Zhao, Chen Liu, Fanyang Mo, Hao Li, Jixiang Zhao, Kaiqing Lin, Liuzhenghao Lv, Li Yuan, Shanzhuo Zhang, Yimi Wang, Zhiyuan Yan","submitted_at":"2026-05-15T17:44:27Z","abstract_excerpt":"Molecules are graphs, but large language models~(LLMs) are usually asked to reason about them through linear strings. The most popular molecular representation, SMILES, compresses atoms, bonds, branches and rings into a compact sequence in which topology is implicit, forcing LLMs to reconstruct molecular structure before performing the requested chemical operation. Here we introduce MoleCode, an LLM-native, training-free, graph-explicit molecular language in which all molecular components are represented as typed entities with persistent identifiers and explicit relations. MoleCode makes molec"},"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":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2605.16480","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/publicdomain/zero/1.0/","primary_cat":"q-bio.BM","submitted_at":"2026-05-15T17:44:27Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"94ca4ad630776d953f42e36b87e77b7d7f40f80707668682182271bc1a939dae","abstract_canon_sha256":"78e6f7213125c1641475524bffea4e4df0e068cb519dbc53f2b99d300786c270"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:02:24.235821Z","signature_b64":"4oaadsYozY7FjMAn/LtXXSYgk8ALt+/YoTvn0+F7vdBddGjrMnAq0TdP3JWsatWsuc95zQRZXgkOeTI9zgtHBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"025cfd2d8ee75245af7ae166727ef6a0ff5287b2feb8be4d271e69c004e7c266","last_reissued_at":"2026-05-20T00:02:24.235032Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:02:24.235032Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MoleCode unlocks structural intelligence in large language models","license":"http://creativecommons.org/publicdomain/zero/1.0/","headline":"MoleCode makes molecular topology directly readable, editable and auditable by LLMs instead of hidden in SMILES strings.","cross_cats":["cs.AI"],"primary_cat":"q-bio.BM","authors_text":"Boxuan Zhao, Chen Liu, Fanyang Mo, Hao Li, Jixiang Zhao, Kaiqing Lin, Liuzhenghao Lv, Li Yuan, Shanzhuo Zhang, Yimi Wang, Zhiyuan Yan","submitted_at":"2026-05-15T17:44:27Z","abstract_excerpt":"Molecules are graphs, but large language models~(LLMs) are usually asked to reason about them through linear strings. The most popular molecular representation, SMILES, compresses atoms, bonds, branches and rings into a compact sequence in which topology is implicit, forcing LLMs to reconstruct molecular structure before performing the requested chemical operation. Here we introduce MoleCode, an LLM-native, training-free, graph-explicit molecular language in which all molecular components are represented as typed entities with persistent identifiers and explicit relations. MoleCode makes molec"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"MoleCode makes molecular topology directly readable, editable and auditable within the language context, allowing an LLM to operate on structure rather than recover it from syntax.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That frontier LLMs can immediately leverage the explicit Subgraph-Node-Edge grammar in prompts for improved reasoning without any training or fine-tuning, and that observed gains stem specifically from structural access rather than prompt length or other variables.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"MoleCode is a training-free, LLM-native representation that makes molecular graphs with explicit atoms, bonds, and topology directly readable and editable in language models, improving structural tasks over implicit string encodings.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"MoleCode makes molecular topology directly readable, editable and auditable by LLMs instead of hidden in SMILES strings.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"9f53fa8e24ecd9989e7fe0994890f60441482db045e537b4ec064c87db3b403d"},"source":{"id":"2605.16480","kind":"arxiv","version":1},"verdict":{"id":"21a50d9b-10d0-41ec-942f-8806fee0b0e7","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T21:34:51.396221Z","strongest_claim":"MoleCode makes molecular topology directly readable, editable and auditable within the language context, allowing an LLM to operate on structure rather than recover it from syntax.","one_line_summary":"MoleCode is a training-free, LLM-native representation that makes molecular graphs with explicit atoms, bonds, and topology directly readable and editable in language models, improving structural tasks over implicit string encodings.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That frontier LLMs can immediately leverage the explicit Subgraph-Node-Edge grammar in prompts for improved reasoning without any training or fine-tuning, and that observed gains stem specifically from structural access rather than prompt length or other variables.","pith_extraction_headline":"MoleCode makes molecular topology directly readable, editable and auditable by LLMs instead of hidden in SMILES strings."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.16480/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_title_agreement","ran_at":"2026-05-19T22:01:23.215207Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T21:41:21.916494Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T19:33:23.110981Z","status":"skipped","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T19:21:57.035161Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"e98367529e4c0a414fbde23364b7059aa05f8bde301d7cdee88f9601d6f726a8"},"references":{"count":60,"sample":[{"doi":"","year":2025,"title":"A survey on large language models in biology and chemistry","work_id":"68571fc3-630e-475f-80bf-3a40961495f7","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Large language models as molecular design engines","work_id":"6199e2c8-591e-4370-a725-a59c08c8c816","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Llamo: Large language model-based molecular graph assistant","work_id":"4976c0e1-a8ba-4e78-b11e-60b60e5c996f","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"A framework for evaluating the chemical knowledge and reasoning abilities of large language models against the expertise of chemists","work_id":"355cd5f1-c18c-4253-8456-ab50af239723","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"arXiv preprint arXiv:2204.11817 , year=","work_id":"8387d3c7-afc3-49b0-ab16-b7f7898f2645","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":60,"snapshot_sha256":"032c0af53b7a6b2466ac92ee4a07767ae3f9909cf015a478c7160b404f29de97","internal_anchors":2},"formal_canon":{"evidence_count":2,"snapshot_sha256":"f872f4c400b8d3c6e36b52a00555ac81de340c061d666c6aa6c8ba6ba669ac2b"},"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":"2605.16480","created_at":"2026-05-20T00:02:24.235154+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.16480v1","created_at":"2026-05-20T00:02:24.235154+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.16480","created_at":"2026-05-20T00:02:24.235154+00:00"},{"alias_kind":"pith_short_12","alias_value":"AJOP2LMO45JE","created_at":"2026-05-20T00:02:24.235154+00:00"},{"alias_kind":"pith_short_16","alias_value":"AJOP2LMO45JELL32","created_at":"2026-05-20T00:02:24.235154+00:00"},{"alias_kind":"pith_short_8","alias_value":"AJOP2LMO","created_at":"2026-05-20T00:02:24.235154+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/AJOP2LMO45JELL324FTHE7XWUD","json":"https://pith.science/pith/AJOP2LMO45JELL324FTHE7XWUD.json","graph_json":"https://pith.science/api/pith-number/AJOP2LMO45JELL324FTHE7XWUD/graph.json","events_json":"https://pith.science/api/pith-number/AJOP2LMO45JELL324FTHE7XWUD/events.json","paper":"https://pith.science/paper/AJOP2LMO"},"agent_actions":{"view_html":"https://pith.science/pith/AJOP2LMO45JELL324FTHE7XWUD","download_json":"https://pith.science/pith/AJOP2LMO45JELL324FTHE7XWUD.json","view_paper":"https://pith.science/paper/AJOP2LMO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.16480&json=true","fetch_graph":"https://pith.science/api/pith-number/AJOP2LMO45JELL324FTHE7XWUD/graph.json","fetch_events":"https://pith.science/api/pith-number/AJOP2LMO45JELL324FTHE7XWUD/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/AJOP2LMO45JELL324FTHE7XWUD/action/timestamp_anchor","attest_storage":"https://pith.science/pith/AJOP2LMO45JELL324FTHE7XWUD/action/storage_attestation","attest_author":"https://pith.science/pith/AJOP2LMO45JELL324FTHE7XWUD/action/author_attestation","sign_citation":"https://pith.science/pith/AJOP2LMO45JELL324FTHE7XWUD/action/citation_signature","submit_replication":"https://pith.science/pith/AJOP2LMO45JELL324FTHE7XWUD/action/replication_record"}},"created_at":"2026-05-20T00:02:24.235154+00:00","updated_at":"2026-05-20T00:02:24.235154+00:00"}