{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.12821", "kind": "paper", "url": "https://arxiv.org/abs/2606.12821"}], "authors": ["Gabriel Diaz-Ireland", "Diego Prieto-Herráez", "Mario García Peces", "Javier Velázquez", "Devika Jain"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.12821", "source_url": "https://arxiv.org/abs/2606.12821"}, "categories": ["General AI", "Language & Knowledge", "cs.AI"], "description": {"en": "GeoNatureAgent Benchmark evaluates LLM agents on environmental geospatial analysis through structured tool calls to a self-hostable API. It includes 93 tasks across 18 categories, with metrics for accuracy and cost."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.12821", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.12821", "title": "GeoNatureAgent Benchmark"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Johns Hopkins University", "Catholic University of Ávila", "Harvard University"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2606.12821", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2606.12821", "source_url": "https://openalex.org/W7164543465"}, "key": "claire-radar:2606.12821", "modality": null, "name": "GeoNatureAgent Benchmark", "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_geonatureagent-benchmark_854d9ae1", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "544399673cdbb555f5bc83be869998a88d37741d0a6bc5c7a346c8c5a2640560", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.12821"}, "publisher": null, "released": "2026-06-11", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.12821", "source_url": "https://arxiv.org/abs/2606.12821"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-12821", "source": "claire_radar", "source_benchmark_id": "2606.12821", "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": null, "hfPaperUrl": "https://huggingface.co/papers/2606.12821", "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": "258c8d828d24cdc11b7d02349ed26a10fff958e2f5deea883d5a0db021929c5a", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "GeoNatureAgent Benchmark evaluates LLM agents on environmental geospatial analysis through structured tool calls to a self-hostable API. It includes 93 tasks across 18 categories, with metrics for accuracy and cost.", "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 the GeoNatureAgent Benchmark, the first benchmark for environmental analysis agents that operate via structured tool calls to a production-style geospatial API."}, "familyId": "bmf_245d2d78860d", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_geonatureagent-benchmark_854d9ae1", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2606.12821", "paper": "https://arxiv.org/abs/2606.12821", "pdf": "https://arxiv.org/pdf/2606.12821", "project": null, "report": "https://arxiv.org/abs/2606.12821"}, "metrics": [], "motivation": "Environmental scientists spend disproportionate effort on data wrangling rather than analysis, and AI agents that automate geospatial workflows remain unvalidated: no benchmark evaluates agents operating through structured tool calling against real APIs.", "name": "GeoNatureAgent Benchmark", "oneLine": "GeoNatureAgent Benchmark evaluates LLM agents on environmental geospatial analysis through structured tool calls to a self-hostable API. It includes 93 tasks across 18 categories, with metrics for accuracy and cost.", "paperTitle": "GeoNatureAgent Benchmark: Benchmarking LLM Agents for Environmental Geospatial Analysis Across Frontier and Open-Weight Foundation Models", "primaryDomain": "General AI", "publication": {"evidence": "The 34th ACM International Conference on Advances in Geographic Information Systems (SIGSPATIAL '26), November 03-06, 2026, Riverside, CA, USA", "evidenceLevel": "strong-author-metadata", "evidenceUrl": "https://arxiv.org/abs/2606.12821", "source": "arxiv-journal-reference", "status": "publication_reported", "venue": "The 34th ACM International Conference on Advances in Geographic Information Systems (SIGSPATIAL '26), November 03-06, 2026, Riverside, CA, USA", "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [{"evidence": [{"level": "strong-author-metadata", "observedAt": "2026-10-14T01:46:58.058199Z", "rawValue": "The 34th ACM International Conference on Advances in Geographic Information Systems (SIGSPATIAL '26), November 03-06, 2026, Riverside, CA, USA", "sourceType": "arxiv-journal-reference", "sourceUrl": "https://arxiv.org/abs/2606.12821"}], "publicationStatus": "published", "venueName": "The 34th ACM International Conference on Advances in Geographic Information Systems (SIGSPATIAL '26), November 03-06, 2026, Riverside, CA, USA"}], "ranking": {}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-06-11", "source": {"authors": ["Gabriel Diaz-Ireland", "Diego Prieto-Herr\\'aez", "Mario Garc\\'ia Peces", "Javier Vel\\'azquez", "Devika Jain"], "categories": ["cs.AI", "cs.ET"], "id": "2606.12821", "title": "GeoNatureAgent Benchmark: Benchmarking LLM Agents for Environmental Geospatial Analysis Across Frontier and Open-Weight Foundation Models", "type": "arxiv", "url": "https://arxiv.org/abs/2606.12821"}, "sourceUpdatedAt": "2026-06-11T02:35:20Z", "topics": ["cs.AI"], "venueAttempts": [], "whyItMatters": "There is a lack of benchmarks for agentic geospatial workflows, and this benchmark provides a realistic API-based environment, enabling assessment of tool-use reasoning and cost-efficiency trade-offs."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
