{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.16802", "kind": "paper", "url": "https://arxiv.org/abs/2606.16802"}], "authors": ["Anqi Zou", "Han Deng", "Chengyu Zhang", "Junquan Hu", "Yu Wang", "Yuxiang Xing", "Aokai Zhang", "Hanling Zhang", "Zhaoyang Liu", "Ben Fei", "Zhihui Wang", "Wanli Ouyang"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.16802", "source_url": "https://arxiv.org/abs/2606.16802"}, "categories": ["Agents & Tool Use", "General AI", "Multimodal"], "description": {"en": "LabOSBench evaluates multimodal GUI agents on 96 subtasks across eight web-based scientific-instrument simulators, covering workflows from sample loading to result inspection. Agents operate via a browser, with execution-based evaluation on task completion."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.16802", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.16802", "title": "LabOSBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2606.16802", "modality": null, "name": "LabOSBench", "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_labosbench_5eb589a2", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "3ef7a42b8d51cd5f174c68935d8de01b032cde7852a0573a0762c3fcd48de98f", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.16802"}, "publisher": null, "released": "2026-06-15", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.16802", "source_url": "https://arxiv.org/abs/2606.16802"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-16802", "source": "claire_radar", "source_benchmark_id": "2606.16802", "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": null, "githubScope": null, "githubStars": null, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2606.16802", "observedAt": "2026-10-04T03:02:42.754940Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "state": "unavailable"}}, "status": "unavailable"}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "ddc0246d344550a2396fe637254122572a12cb621d64e6d5f4c3f52ffdb7aac5", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "LabOSBench evaluates multimodal GUI agents on 96 subtasks across eight web-based scientific-instrument simulators, covering workflows from sample loading to result inspection. Agents operate via a browser, with execution-based evaluation on task completion.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "To this end, we introduce LabOSBench, a challenging benchmark for multimodal GUI agents built on a suite of web-based scientific-instrument simulators."}, "familyId": "bmf_89ea60d8b827", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_labosbench_5eb589a2", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2606.16802", "paper": "https://arxiv.org/abs/2606.16802", "pdf": "https://arxiv.org/pdf/2606.16802", "project": null, "report": "https://arxiv.org/abs/2606.16802"}, "metrics": [], "motivation": "Current computer-use benchmarks primarily focus on software operation tasks in virtualized systems, whereas scientific instrumentation scenarios require coordinated control over complex interfaces, and feedback-driven parameter adjustment.", "name": "LabOSBench", "oneLine": "LabOSBench evaluates multimodal GUI agents on 96 subtasks across eight web-based scientific-instrument simulators, covering workflows from sample loading to result inspection. Agents operate via a browser, with execution-based evaluation on task completion.", "paperTitle": "LabOSBench: Benchmarking Computer Use Agents for Scientific Instrument Control", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2606.16802", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "arXiv", "organizationType": "benchmark-organization", "role": "benchmark-publisher", "sourceUrl": "https://arxiv.org/abs/2606.16802"}], "ranking": {}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-06-15", "source": {"authors": ["Anqi Zou", "Han Deng", "Chengyu Zhang", "Junquan Hu", "Yu Wang", "Yuxiang Xing", "Aokai Zhang", "Hanling Zhang", "Zhaoyang Liu", "Ben Fei", "Zhihui Wang and Wanli Ouyang"], "categories": ["cs.AI"], "id": "2606.16802", "title": "LabOSBench: Benchmarking Computer Use Agents for Scientific Instrument Control", "type": "arxiv", "url": "https://arxiv.org/abs/2606.16802"}, "sourceUpdatedAt": "2026-06-15T14:42:33Z", "topics": ["Multimodal"], "venueAttempts": [], "whyItMatters": "Existing computer-use benchmarks focus on software tasks, leaving a gap for scientific instrument control. LabOSBench provides a safe, reproducible, low-cost testbed to assess agents' feedback-driven and long-horizon capabilities in instrument operation, supporting practical adoption in laboratory automation."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
