{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2607.06008", "kind": "paper", "url": "https://arxiv.org/abs/2607.06008"}], "authors": ["Hongliang Li", "Yijin Liu", "Zhiwei Zhang", "Zihe Liu", "Xinyue Lou", "Jinan Xu", "Fandong Meng", "Kaiyu Huang"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2607.06008", "source_url": "https://arxiv.org/abs/2607.06008"}, "categories": ["General AI", "Language & Knowledge", "cs.AI"], "description": {"en": "PolyWorkBench evaluates LLM agents on multilingual, long-horizon workplace workflows across five domains: commerce, knowledge work, legal analysis, localization, and manufacturing. It includes 67 tasks, structured scoring via Grade, executable state verification with Pytest, and LLM-as-Judge diagnostics."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2607.06008", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2607.06008", "title": "PolyWorkBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Beijing Jiaotong University", "Tencent (China)"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2607.06008", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2607.06008", "source_url": "https://openalex.org/W7167739024"}, "key": "claire-radar:2607.06008", "modality": null, "name": "PolyWorkBench", "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_polyworkbench_2abf54e5", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "f711b939f5832e463aed08349ac766cf91986fd74687d42ca7501a1bc794d737", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2607.06008"}, "publisher": null, "released": "2026-07-07", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2607.06008", "source_url": "https://arxiv.org/abs/2607.06008"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2607-06008", "source": "claire_radar", "source_benchmark_id": "2607.06008", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Language & Knowledge", "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": 0, "hfPaperUrl": "https://huggingface.co/papers/2607.06008", "observedAt": "2026-10-04T15:31:48.030916Z", "signalStatus": {"githubStars": {"state": "not_applicable"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}}, "status": "fresh"}, "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": "40e85edc51a0543817b5765dc2b8757fba2da41c8f8d09fa6b86e7a1e309dcee", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "PolyWorkBench evaluates LLM agents on multilingual, long-horizon workplace workflows across five domains: commerce, knowledge work, legal analysis, localization, and manufacturing. It includes 67 tasks, structured scoring via Grade, executable state verification with Pytest, and LLM-as-Judge diagnostics.", "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": "We introduce PolyWorkBench, a benchmark designed to evaluate LLM agents on multilingual, long-horizon workplace workflows."}, "familyId": "bmf_91789172102b", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_polyworkbench_2abf54e5", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2607.06008", "paper": "https://arxiv.org/abs/2607.06008", "pdf": "https://arxiv.org/pdf/2607.06008", "project": null, "report": "https://arxiv.org/abs/2607.06008"}, "metrics": [], "motivation": "While Large Language Model (LLM) agents excel at monolingual long-horizon planning and tool use, enterprise workflows inherently require processing multilingual resources across extended trajectories.", "name": "PolyWorkBench", "oneLine": "PolyWorkBench evaluates LLM agents on multilingual, long-horizon workplace workflows across five domains: commerce, knowledge work, legal analysis, localization, and manufacturing. It includes 67 tasks, structured scoring via Grade, executable state verification with Pytest, and LLM-as-Judge diagnostics.", "paperTitle": "PolyWorkBench: Benchmarking LLM Agents for Cross-Lingual Long-Horizon Workflows", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2607.06008", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "PolyWorkBench Team", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://arxiv.org/abs/2607.06008"}], "ranking": {"90d": {"confidence": "Low", "coverage": 0.15, "growth": {"components": {"githubStars": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfDatasetDownloads": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfPaperUpvotes": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 90}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": null, "value": null}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.0, "value": 0}}, "confidence": "Low", "coverage": 0.15, "rank": null, "score": 43}, "method": "attention-ranking-v13", "rank": null, "score": 43}}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-07-07", "source": {"authors": ["Hongliang Li", "Yijin Liu", "Zhiwei Zhang", "Zihe Liu", "Xinyue Lou", "Jinan Xu", "Fandong Meng", "Kaiyu Huang"], "categories": ["cs.AI", "cs.CL"], "id": "2607.06008", "title": "PolyWorkBench: Benchmarking LLM Agents for Cross-Lingual Long-Horizon Workflows", "type": "arxiv", "url": "https://arxiv.org/abs/2607.06008"}, "sourceUpdatedAt": "2026-08-17T05:33:41Z", "topics": ["cs.AI"], "venueAttempts": [], "whyItMatters": "This benchmark fills the gap of evaluating LLM agents on tasks that combine multilinguality and long-horizon execution, providing a structured protocol to compare agent performance and identify systematic failure modes in cross-lingual scenarios."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
