{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:E44EJ52FFPTOVF7SSLFGRS6I3G","short_pith_number":"pith:E44EJ52F","schema_version":"1.0","canonical_sha256":"273844f7452be6ea97f292ca68cbc8d9865cc7876d1b06fbd4154330e8b8ec2b","source":{"kind":"arxiv","id":"2507.16572","version":1},"attestation_state":"computed","paper":{"title":"Pixels to Principles: Probing Intuitive Physics Understanding in Multimodal Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Elia Bruni, Mohamad Ballout, Serwan Jassim","submitted_at":"2025-07-22T13:24:42Z","abstract_excerpt":"This paper presents a systematic evaluation of state-of-the-art multimodal large language models (MLLMs) on intuitive physics tasks using the GRASP and IntPhys 2 datasets. We assess the open-source models InternVL 2.5, Qwen 2.5 VL, LLaVA-OneVision, and the proprietary Gemini 2.0 Flash Thinking, finding that even the latest models struggle to reliably distinguish physically plausible from implausible scenarios. To go beyond performance metrics, we conduct a probing analysis of model embeddings, extracting intermediate representations at key processing stages to examine how well task-relevant in"},"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":"2507.16572","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-07-22T13:24:42Z","cross_cats_sorted":[],"title_canon_sha256":"0195dc4e44ff6a6cbbf3ba3c6e27f46cacd8a3efddb039d441bb8e9eaf3d2a5f","abstract_canon_sha256":"507da66bf032375646cbb6ed12726889bb0e395c8664ab26c64c0d611f59cf45"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:41:12.327901Z","signature_b64":"t14BhDSDQv05noGzxr43ZWWE8vzdIM/obIlZ/nNpdEIDaJ6Fw8F+USKyvi/1ZDPAi3o6E9xYBWpi/AvG3JMgDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"273844f7452be6ea97f292ca68cbc8d9865cc7876d1b06fbd4154330e8b8ec2b","last_reissued_at":"2026-07-05T11:41:12.327444Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:41:12.327444Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Pixels to Principles: Probing Intuitive Physics Understanding in Multimodal Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Elia Bruni, Mohamad Ballout, Serwan Jassim","submitted_at":"2025-07-22T13:24:42Z","abstract_excerpt":"This paper presents a systematic evaluation of state-of-the-art multimodal large language models (MLLMs) on intuitive physics tasks using the GRASP and IntPhys 2 datasets. We assess the open-source models InternVL 2.5, Qwen 2.5 VL, LLaVA-OneVision, and the proprietary Gemini 2.0 Flash Thinking, finding that even the latest models struggle to reliably distinguish physically plausible from implausible scenarios. To go beyond performance metrics, we conduct a probing analysis of model embeddings, extracting intermediate representations at key processing stages to examine how well task-relevant in"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.16572","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/2507.16572/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":"2507.16572","created_at":"2026-07-05T11:41:12.327501+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.16572v1","created_at":"2026-07-05T11:41:12.327501+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.16572","created_at":"2026-07-05T11:41:12.327501+00:00"},{"alias_kind":"pith_short_12","alias_value":"E44EJ52FFPTO","created_at":"2026-07-05T11:41:12.327501+00:00"},{"alias_kind":"pith_short_16","alias_value":"E44EJ52FFPTOVF7S","created_at":"2026-07-05T11:41:12.327501+00:00"},{"alias_kind":"pith_short_8","alias_value":"E44EJ52F","created_at":"2026-07-05T11:41:12.327501+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/E44EJ52FFPTOVF7SSLFGRS6I3G","json":"https://pith.science/pith/E44EJ52FFPTOVF7SSLFGRS6I3G.json","graph_json":"https://pith.science/api/pith-number/E44EJ52FFPTOVF7SSLFGRS6I3G/graph.json","events_json":"https://pith.science/api/pith-number/E44EJ52FFPTOVF7SSLFGRS6I3G/events.json","paper":"https://pith.science/paper/E44EJ52F"},"agent_actions":{"view_html":"https://pith.science/pith/E44EJ52FFPTOVF7SSLFGRS6I3G","download_json":"https://pith.science/pith/E44EJ52FFPTOVF7SSLFGRS6I3G.json","view_paper":"https://pith.science/paper/E44EJ52F","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.16572&json=true","fetch_graph":"https://pith.science/api/pith-number/E44EJ52FFPTOVF7SSLFGRS6I3G/graph.json","fetch_events":"https://pith.science/api/pith-number/E44EJ52FFPTOVF7SSLFGRS6I3G/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/E44EJ52FFPTOVF7SSLFGRS6I3G/action/timestamp_anchor","attest_storage":"https://pith.science/pith/E44EJ52FFPTOVF7SSLFGRS6I3G/action/storage_attestation","attest_author":"https://pith.science/pith/E44EJ52FFPTOVF7SSLFGRS6I3G/action/author_attestation","sign_citation":"https://pith.science/pith/E44EJ52FFPTOVF7SSLFGRS6I3G/action/citation_signature","submit_replication":"https://pith.science/pith/E44EJ52FFPTOVF7SSLFGRS6I3G/action/replication_record"}},"created_at":"2026-07-05T11:41:12.327501+00:00","updated_at":"2026-07-05T11:41:12.327501+00:00"}