{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:DMNDIYZ6LD3KV44VFERAOOUZQY","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"7111c05fde22dd9f75c971e49ca175237b4901d279b78cd95c37ba1faa83da92","cross_cats_sorted":["cs.CV"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-03-27T08:10:05Z","title_canon_sha256":"643eafe4d72694bb367d0827179831d9525fd75bafb8ce520b2208c13252d20c"},"schema_version":"1.0","source":{"id":"2603.26839","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2603.26839","created_at":"2026-05-18T03:09:22Z"},{"alias_kind":"arxiv_version","alias_value":"2603.26839v2","created_at":"2026-05-18T03:09:22Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.26839","created_at":"2026-05-18T03:09:22Z"},{"alias_kind":"pith_short_12","alias_value":"DMNDIYZ6LD3K","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"DMNDIYZ6LD3KV44V","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"DMNDIYZ6","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:7ae046e67b2ac5abb8a45552cd1437a89500e8d42a35d2369494f2e65c0f893a","target":"graph","created_at":"2026-05-18T03:09:22Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"MazeBench therefore shows that high accuracy on visual planning tasks does not imply human-like spatial understanding."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the two-stage image-to-grid plus token enumeration strategy observed in traces is the dominant mechanism driving performance rather than an artifact of the specific prompting or model configurations tested."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Multimodal models achieve high maze accuracy by translating images to text grids and performing token-level BFS search, not through visual planning."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Multimodal models solve maze images by converting them to text grids and enumerating paths token by token rather than through visual planning."}],"snapshot_sha256":"d976b97cdf972bb6cafa6e406e76c79ea9bf3feaa696102766660ef5b8dbfcb3"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"2fe155cd7c9677a14573ccbf4acbc8106e30e888fad95d92065dbff341312706"},"paper":{"abstract_excerpt":"How do multimodal models solve visual spatial tasks -- through genuine planning, or through brute-force search in token space? We introduce \\textsc{MazeBench}, a benchmark of 110 procedurally generated maze images across nine controlled groups, and evaluate 16 model configurations from OpenAI, Anthropic, Google, and Alibaba. GPT-5.4 solves 91\\% and Gemini 3.1 Pro 79\\%, but these scores are misleading: models typically translate images into text grids and then enumerate paths step by step, consuming 1,710--22,818 tokens per solve for a task humans do quickly. Without added reasoning budgets, al","authors_text":"Alberto G. Rodriguez Salgado","cross_cats":["cs.CV"],"headline":"Multimodal models solve maze images by converting them to text grids and enumerating paths token by token rather than through visual planning.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-03-27T08:10:05Z","title":"From Pixels to BFS: High Maze Accuracy Does Not Imply Visual Planning"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2603.26839","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-15T00:05:18.200340Z","id":"bf6d08d8-4812-4267-97ea-c988b94b243a","model_set":{"reader":"grok-4.3"},"one_line_summary":"Multimodal models achieve high maze accuracy by translating images to text grids and performing token-level BFS search, not through visual planning.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Multimodal models solve maze images by converting them to text grids and enumerating paths token by token rather than through visual planning.","strongest_claim":"MazeBench therefore shows that high accuracy on visual planning tasks does not imply human-like spatial understanding.","weakest_assumption":"That the two-stage image-to-grid plus token enumeration strategy observed in traces is the dominant mechanism driving performance rather than an artifact of the specific prompting or model configurations tested."}},"verdict_id":"bf6d08d8-4812-4267-97ea-c988b94b243a"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:ab0f6d0407c38f74b8d1848efd8348568f1126575a58bfcf534830d9d564ca35","target":"record","created_at":"2026-05-18T03:09:22Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"7111c05fde22dd9f75c971e49ca175237b4901d279b78cd95c37ba1faa83da92","cross_cats_sorted":["cs.CV"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-03-27T08:10:05Z","title_canon_sha256":"643eafe4d72694bb367d0827179831d9525fd75bafb8ce520b2208c13252d20c"},"schema_version":"1.0","source":{"id":"2603.26839","kind":"arxiv","version":2}},"canonical_sha256":"1b1a34633e58f6aaf3952922073a99861b264bfad4e1494fc56067e07e3f0740","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"1b1a34633e58f6aaf3952922073a99861b264bfad4e1494fc56067e07e3f0740","first_computed_at":"2026-05-18T03:09:22.499408Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T03:09:22.499408Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"TDW5+AcfaGAmLsOzz8ScRSejGV6ZXtzKgX++IpIGjHRU5+yQ02jKDiQDY1F5DMRbH4AZ0j4u2UyUD3bVocGUAA==","signature_status":"signed_v1","signed_at":"2026-05-18T03:09:22.499994Z","signed_message":"canonical_sha256_bytes"},"source_id":"2603.26839","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:ab0f6d0407c38f74b8d1848efd8348568f1126575a58bfcf534830d9d564ca35","sha256:7ae046e67b2ac5abb8a45552cd1437a89500e8d42a35d2369494f2e65c0f893a"],"state_sha256":"47b48deb43abaa93131cb93337468ea282f697f80911b3ceb6c1bace26029697"}