{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2607.19261", "kind": "paper", "url": "https://arxiv.org/abs/2607.19261"}], "authors": ["Dankai Liao", "Tianyi Zhang", "Yufeng Wu", "Xinyue Zhang", "Qiaochu Xue", "Zeyu Liu", "Dachun Zhao", "Linghan Cai", "Yueming Jin"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2607.19261", "source_url": "https://arxiv.org/abs/2607.19261"}, "categories": ["Biology & Drug Discovery", "Information retrieval", "Multimodal", "Reasoning"], "description": {"en": "PathAgentBench evaluates vision-language models on whole-slide pathology images across four capabilities: image-to-text matching, text-to-image retrieval, diagnostic-region localization, and multi-scale reasoning. It includes 1,822 TCGA WSIs and 17,135 diagnostic paths."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2607.19261", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2607.19261", "title": "PathAgentBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2607.19261", "modality": null, "name": "PathAgentBench", "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_pathagentbench_70491a6b", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "41b3d475e9a4536932c466d9bf1c98425de534c620d776ae857e252ac5e2978b", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2607.19261"}, "publisher": null, "released": "2026-07-21", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2607.19261", "source_url": "https://arxiv.org/abs/2607.19261"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2607-19261", "source": "claire_radar", "source_benchmark_id": "2607.19261", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["Biology & Drug Discovery"], "area": "Multimodal", "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/2607.19261", "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": ["Reasoning", "Information retrieval"], "confidence": "High", "construction": "Transform Existing", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "164a62c1666c683614d9776c3857d004743532095c87209c9cd9c0e1c75bad41", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "PathAgentBench evaluates vision-language models on whole-slide pathology images across four capabilities: image-to-text matching, text-to-image retrieval, diagnostic-region localization, and multi-scale reasoning. It includes 1,822 TCGA WSIs and 17,135 diagnostic paths.", "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 PathAgentBench, a benchmark for evaluating evidence-seeking vision-language models (VLMs) across four complementary capabilities: image-to-text matching for evidence interpretation, text-to-image retrieval for evidence verification, diagnostic-region localization for evidence acquisition, and multi-scale reasoning for evidence integration."}, "familyId": "bmf_70918a81327d", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_pathagentbench_70491a6b", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": ["Pharma & Biotech"], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2607.19261", "paper": "https://arxiv.org/abs/2607.19261", "pdf": "https://arxiv.org/pdf/2607.19261", "project": null, "report": "https://arxiv.org/abs/2607.19261"}, "metrics": [], "motivation": "Whole-slide image (WSI) diagnosis requires identifying diagnostically relevant regions, examining them across magnifications, and integrating multi-scale evidence.", "name": "PathAgentBench", "oneLine": "PathAgentBench evaluates vision-language models on whole-slide pathology images across four capabilities: image-to-text matching, text-to-image retrieval, diagnostic-region localization, and multi-scale reasoning. It includes 1,822 TCGA WSIs and 17,135 diagnostic paths.", "paperTitle": "PathAgentBench: Benchmarking Evidence-Seeking Vision-Language Models on Whole-Slide Pathology Image", "primaryDomain": "Biology & Drug Discovery", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2607.19261", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "ranking": {"90d": {"confidence": "Low", "coverage": 0.0, "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": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "rank": null, "score": null}, "method": "attention-ranking-v13", "rank": null, "score": null}}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-07-21", "source": {"authors": ["Dankai Liao", "Tianyi Zhang", "Yufeng Wu", "Xinyue Zhang", "Qiaochu Xue", "Zeyu Liu", "Dachun Zhao", "Linghan Cai", "and Yueming Jin"], "categories": ["cs.CV", "cs.AI"], "id": "2607.19261", "title": "PathAgentBench: Benchmarking Evidence-Seeking Vision-Language Models on Whole-Slide Pathology Image", "type": "arxiv", "url": "https://arxiv.org/abs/2607.19261"}, "sourceUpdatedAt": "2026-07-30T09:25:28Z", "topics": ["Multimodal", "Reasoning"], "venueAttempts": [], "whyItMatters": "Most pathology benchmarks use pre-cropped patches, not whole-slide exploration. PathAgentBench provides a unified framework with annotated paths, revealing a significant gap in evidence acquisition and supporting progress in autonomous WSI diagnosis."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
