{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:2X4D6OBM6YH5IUXKXFPKLNFTCJ","short_pith_number":"pith:2X4D6OBM","schema_version":"1.0","canonical_sha256":"d5f83f382cf60fd452eab95ea5b4b3126200e122a97732ef136e6d3b3f0a367f","source":{"kind":"arxiv","id":"2607.00501","version":1},"attestation_state":"computed","paper":{"title":"BaseRT: Best-in-Class LLM Inference on Apple Silicon via Native Metal","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.PF"],"primary_cat":"cs.CL","authors_text":"Fabian Waschkowski, Lukas Wesemann, Prabod Rathnayaka","submitted_at":"2026-07-01T06:37:11Z","abstract_excerpt":"We present BaseRT, a native Metal inference runtime for large language models (LLMs) on Apple Silicon, and report the highest inference throughput on this hardware to date. Existing runtimes, including llama.cpp and MLX-based frameworks, incur overhead from abstractions not designed for Metal's execution model or Apple Silicon's unified memory topology. By building natively on Metal with chip-specific kernel fusion, unified memory-aware optimisation, and custom dispatch logic, BaseRT recovers performance that framework-based approaches leave on the table. BaseRT supports a wide range of model "},"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":"2607.00501","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2026-07-01T06:37:11Z","cross_cats_sorted":["cs.AI","cs.PF"],"title_canon_sha256":"0dc6b9db8c364cc1b15be67b8b90b244ba1bf11ea5e5d051172245a398a45e12","abstract_canon_sha256":"058973f910e1fcd4a0596906f30b4c0d1ccfa2999eaa9f485a22577b4fbc47ba"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-02T01:17:45.284220Z","signature_b64":"c46Y28JvTNM1CFIGBYlZ6WzGpkOr48n2IDhwj0K8V7O/CIu1creHAsJJbocyBuQ0G1QCDa3bYFddpMerTtMVDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d5f83f382cf60fd452eab95ea5b4b3126200e122a97732ef136e6d3b3f0a367f","last_reissued_at":"2026-07-02T01:17:45.283821Z","signature_status":"signed_v1","first_computed_at":"2026-07-02T01:17:45.283821Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"BaseRT: Best-in-Class LLM Inference on Apple Silicon via Native Metal","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.PF"],"primary_cat":"cs.CL","authors_text":"Fabian Waschkowski, Lukas Wesemann, Prabod Rathnayaka","submitted_at":"2026-07-01T06:37:11Z","abstract_excerpt":"We present BaseRT, a native Metal inference runtime for large language models (LLMs) on Apple Silicon, and report the highest inference throughput on this hardware to date. Existing runtimes, including llama.cpp and MLX-based frameworks, incur overhead from abstractions not designed for Metal's execution model or Apple Silicon's unified memory topology. By building natively on Metal with chip-specific kernel fusion, unified memory-aware optimisation, and custom dispatch logic, BaseRT recovers performance that framework-based approaches leave on the table. BaseRT supports a wide range of model "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.00501","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/2607.00501/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":"2607.00501","created_at":"2026-07-02T01:17:45.283878+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.00501v1","created_at":"2026-07-02T01:17:45.283878+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.00501","created_at":"2026-07-02T01:17:45.283878+00:00"},{"alias_kind":"pith_short_12","alias_value":"2X4D6OBM6YH5","created_at":"2026-07-02T01:17:45.283878+00:00"},{"alias_kind":"pith_short_16","alias_value":"2X4D6OBM6YH5IUXK","created_at":"2026-07-02T01:17:45.283878+00:00"},{"alias_kind":"pith_short_8","alias_value":"2X4D6OBM","created_at":"2026-07-02T01:17:45.283878+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/2X4D6OBM6YH5IUXKXFPKLNFTCJ","json":"https://pith.science/pith/2X4D6OBM6YH5IUXKXFPKLNFTCJ.json","graph_json":"https://pith.science/api/pith-number/2X4D6OBM6YH5IUXKXFPKLNFTCJ/graph.json","events_json":"https://pith.science/api/pith-number/2X4D6OBM6YH5IUXKXFPKLNFTCJ/events.json","paper":"https://pith.science/paper/2X4D6OBM"},"agent_actions":{"view_html":"https://pith.science/pith/2X4D6OBM6YH5IUXKXFPKLNFTCJ","download_json":"https://pith.science/pith/2X4D6OBM6YH5IUXKXFPKLNFTCJ.json","view_paper":"https://pith.science/paper/2X4D6OBM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.00501&json=true","fetch_graph":"https://pith.science/api/pith-number/2X4D6OBM6YH5IUXKXFPKLNFTCJ/graph.json","fetch_events":"https://pith.science/api/pith-number/2X4D6OBM6YH5IUXKXFPKLNFTCJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2X4D6OBM6YH5IUXKXFPKLNFTCJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2X4D6OBM6YH5IUXKXFPKLNFTCJ/action/storage_attestation","attest_author":"https://pith.science/pith/2X4D6OBM6YH5IUXKXFPKLNFTCJ/action/author_attestation","sign_citation":"https://pith.science/pith/2X4D6OBM6YH5IUXKXFPKLNFTCJ/action/citation_signature","submit_replication":"https://pith.science/pith/2X4D6OBM6YH5IUXKXFPKLNFTCJ/action/replication_record"}},"created_at":"2026-07-02T01:17:45.283878+00:00","updated_at":"2026-07-02T01:17:45.283878+00:00"}