{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2607.28545", "kind": "paper", "url": "https://arxiv.org/abs/2607.28545"}, {"id": "gh:orca-bench/orca-bench", "kind": "repo", "url": "https://github.com/ORCA-bench/ORCA-bench"}, {"kind": "website", "url": "https://hub.harborframework.com/datasets/orca-bench/orca-bench"}], "authors": ["Albert Gong", "Kyuseong Choi", "Abhineet Agarwal", "Jason Schechner", "Ryan Huang", "Raj Agrawal", "Anish Agarwal", "Raaz Dwivedi"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2607.28545", "source_url": "https://arxiv.org/abs/2607.28545"}, "categories": ["Code & Software", "Software & AI Compute", "cs.CL"], "description": {"en": "ORCA-bench evaluates language model agents on root cause analysis in a production-fidelity oncall setting, using a live OpenTelemetry-instrumented microservice system with 1,079 tasks varying in report specificity, time-to-detection, and fault co-occurrence. Agents access metrics, logs, traces, and source code via real telemetry interfaces."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2607.28545", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2607.28545", "title": "ORCA-bench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Cornell Tech", "Columbia University"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2607.28545", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2607.28545", "source_url": "https://openalex.org/W7171979996"}, "key": "claire-radar:2607.28545", "modality": null, "name": "ORCA-bench", "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_orca-bench_79a9e07b", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "85d74ec90d3b7bd6b1a17af0bd8963ebb1a34f7cbac0c47ba1999cba6c38d216", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2607.28545"}, "publisher": null, "released": "2026-07-30", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2607.28545", "source_url": "https://arxiv.org/abs/2607.28545"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2607-28545", "source": "claire_radar", "source_benchmark_id": "2607.28545", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["Software & AI Compute"], "area": "Code & Software", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/ORCA-bench/ORCA-bench", "githubScope": "benchmark_repo", "githubStars": 5, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2607.28545", "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", "state": "unavailable"}}, "status": "fresh"}, "capabilities": [], "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": "cd8c69fe2003f3b637d6ae1908529e54338a40d5249f2b3e299e0934d6c49976", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "ORCA-bench evaluates language model agents on root cause analysis in a production-fidelity oncall setting, using a live OpenTelemetry-instrumented microservice system with 1,079 tasks varying in report specificity, time-to-detection, and fault co-occurrence. Agents access metrics, logs, traces, and source code via real telemetry interfaces.", "domainCuration": {"method": "primary-source-review-v1", "state": "reviewed"}, "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL"], "snippet": "We introduce ORCA-bench, a benchmark that puts general-purpose coding agents in a production-fidelity oncall setting."}, "familyId": "bmf_ee99b02dd1db", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [{"evidence": [{"basis": "linked-from-project", "linkedUrl": "https://github.com/ORCA-bench/ORCA-bench", "url": "https://hub.harborframework.com/datasets/orca-bench/orca-bench"}], "scope": "benchmark_repo", "url": "https://github.com/ORCA-bench/ORCA-bench", "verification": "source-linked"}], "status": "linked"}, "heat": null, "id": "bm_orca-bench_79a9e07b", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": ["Software & Cloud"], "links": {"code": "https://github.com/ORCA-bench/ORCA-bench", "data": null, "hfPaper": "https://huggingface.co/papers/2607.28545", "paper": "https://arxiv.org/abs/2607.28545", "pdf": "https://arxiv.org/pdf/2607.28545", "project": "https://hub.harborframework.com/datasets/orca-bench/orca-bench", "report": "https://arxiv.org/abs/2607.28545"}, "metrics": [], "motivation": "Large language models can write, patch, and search code, but oncall root cause analysis (RCA) demands something different: reasoning over noisy metrics, logs, traces, and source code, starting from ambiguous user-facing reports, often hours after the incident began.", "name": "ORCA-bench", "oneLine": "ORCA-bench evaluates language model agents on root cause analysis in a production-fidelity oncall setting, using a live OpenTelemetry-instrumented microservice system with 1,079 tasks varying in report specificity, time-to-detection, and fault co-occurrence. Agents access metrics, logs, traces, and source code via real telemetry interfaces.", "paperTitle": "ORCA-bench: How Ready Are Language Model Agents for Oncall?", "primaryDomain": "Software & AI Compute", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2607.28545", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "Harbor Framework", "organizationType": "benchmark-organization", "role": "benchmark-publisher", "sourceUrl": "https://hub.harborframework.com/datasets/orca-bench/orca-bench"}], "ranking": {"90d": {"confidence": "Low", "coverage": 0.55, "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": 0.2391676802567412, "value": 5}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": null, "value": null}}, "confidence": "Low", "coverage": 0.55, "rank": 180, "score": 36}, "method": "attention-ranking-v13", "rank": 180, "score": 36}}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-07-30", "source": {"authors": ["Albert Gong", "Kyuseong Choi", "Abhineet Agarwal", "Jason Schechner", "Ryan Huang", "Raj Agrawal", "Anish Agarwal", "Raaz Dwivedi"], "categories": ["cs.CL", "cs.AI", "cs.SE"], "id": "2607.28545", "title": "ORCA-bench: How Ready Are Language Model Agents for Oncall?", "type": "arxiv", "url": "https://arxiv.org/abs/2607.28545"}, "sourceUpdatedAt": "2026-08-05T12:21:27Z", "topics": ["cs.CL"], "venueAttempts": [], "whyItMatters": "Standard coding benchmarks do not capture the complexity of oncall RCA, where agents must reason over noisy, heterogeneous data. ORCA-bench provides a reproducible testbed to measure agent readiness for production reliability tasks, revealing a significant performance gap even for frontier models."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
