{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.09669", "kind": "paper", "url": "https://arxiv.org/abs/2606.09669"}, {"id": "gh:hongcheng-gao/spatialworld", "kind": "repo", "url": "https://github.com/Hongcheng-Gao/SpatialWorld"}], "authors": ["Hongcheng Gao", "Hailong Qu", "Jingyi Tang", "Jiahao Wang", "Zihao Huang", "Hengkang Qiao", "Shihong Huang", "Junming Yang", "Yi Li", "Hongyixuan Yuan", "Wenjie Li", "Bohan Zeng", "Wenbo Li", "Bo Wang", "Jianhui Liu", "Olive Huang", "Haoyang Huang", "Wentao Zhang", "Guoqing Huang", "Nan Duan", "Yinpeng Dong"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.09669", "source_url": "https://arxiv.org/abs/2606.09669"}, "categories": ["Agents & Tool Use", "General AI", "Multimodal", "Reasoning"], "description": {"en": "SpatialWorld evaluates multimodal agents on interactive spatial reasoning in 760 real-world tasks across eight simulation backends. Agents operate under vision-only partial observability, using a unified text-based action interface. Performance is measured via terminal-state verifiers for task success rate and step efficiency."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.09669", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.09669", "title": "SpatialWorld"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2606.09669", "modality": null, "name": "SpatialWorld", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "High", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "unknown", "origin_record_id": "bm_spatialworld_4f21023e", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "3d976a117a874f919b9d24e0ef8c79265f279f81f80ada70a0b9fab1ea7a4e06", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.09669"}, "publisher": null, "released": "2026-06-08", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.09669", "source_url": "https://arxiv.org/abs/2606.09669"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-09669", "source": "claire_radar", "source_benchmark_id": "2606.09669", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Agents & Tool Use", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/Hongcheng-Gao/SpatialWorld", "githubScope": "benchmark_repo", "githubStars": 55, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 50, "hfPaperUrl": "https://huggingface.co/papers/2606.09669", "observedAt": "2026-10-04T15:31:48.030916Z", "signalStatus": {"githubStars": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "lastSuccessfulDate": "2026-09-25", "state": "stale"}}, "status": "fresh"}, "capabilities": ["Reasoning"], "confidence": "High", "construction": "Interactive Environment", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "0d7ddb4d37209814fd6ecfb0cd6b5bbf2aa8c346230263f46848ee81b8fe5600", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "SpatialWorld evaluates multimodal agents on interactive spatial reasoning in 760 real-world tasks across eight simulation backends. Agents operate under vision-only partial observability, using a unified text-based action interface. Performance is measured via terminal-state verifiers for task success rate and step efficiency.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact coined title identity tied to benchmark evidence", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "We introduce SpatialWorld, a unified benchmark designed specifically for evaluating the interactive spatial understanding of multimodal agents in complex real-world tasks."}, "familyId": "bmf_b45842946201", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [{"evidence": [{"basis": "existing-code-link", "linkedUrl": "https://github.com/Hongcheng-Gao/SpatialWorld", "url": "https://github.com/Hongcheng-Gao/SpatialWorld"}, {"basis": "existing-metric-association", "linkedUrl": "https://github.com/Hongcheng-Gao/SpatialWorld", "url": "https://huggingface.co/papers/2606.09669"}], "scope": "benchmark_repo", "url": "https://github.com/Hongcheng-Gao/SpatialWorld", "verification": "source-linked"}], "status": "linked"}, "heat": null, "id": "bm_spatialworld_4f21023e", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/Hongcheng-Gao/SpatialWorld", "data": null, "hfPaper": "https://huggingface.co/papers/2606.09669", "paper": "https://arxiv.org/abs/2606.09669", "pdf": "https://arxiv.org/pdf/2606.09669", "project": null, "report": "https://arxiv.org/abs/2606.09669"}, "metrics": [], "motivation": "Spatial reasoning is a foundational capability for multimodal large language models (MLLMs) to perceive and operate within the physical world.", "name": "SpatialWorld", "oneLine": "SpatialWorld evaluates multimodal agents on interactive spatial reasoning in 760 real-world tasks across eight simulation backends. Agents operate under vision-only partial observability, using a unified text-based action interface. Performance is measured via terminal-state verifiers for task success rate and step efficiency.", "paperTitle": "SpatialWorld: Benchmarking Interactive Spatial Reasoning of Multimodal Agents in Real-World Tasks", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2606.09669", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "SpatialWorld Team", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://github.com/Hongcheng-Gao/SpatialWorld"}], "ranking": {}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-06-08", "source": {"authors": ["Hongcheng Gao", "Hailong Qu", "Jingyi Tang", "Jiahao Wang", "Zihao Huang", "Hengkang Qiao", "Shihong Huang", "Junming Yang", "Yi Li", "Hongyixuan Yuan", "Wenjie Li", "Bohan Zeng", "Wenbo Li", "Bo Wang", "Jianhui Liu", "Olive Huang", "Haoyang Huang", "Wentao Zhang", "Guoqing Huang", "Nan Duan", "Yinpeng Dong"], "categories": ["cs.AI", "cs.CL"], "id": "2606.09669", "title": "SpatialWorld: Benchmarking Interactive Spatial Reasoning of Multimodal Agents in Real-World Tasks", "type": "arxiv", "url": "https://arxiv.org/abs/2606.09669"}, "sourceUpdatedAt": "2026-06-13T15:55:01Z", "topics": ["Multimodal", "Reasoning"], "venueAttempts": [], "whyItMatters": "Existing benchmarks rely on passive VQA or simulator-specific pipelines, failing to assess interactive spatial understanding. SpatialWorld provides a unified, simulator-agnostic protocol with human-validated evaluation, revealing that current models achieve low success rates, highlighting gaps in active exploration and long-horizon planning."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
