{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.01189", "kind": "paper", "url": "https://arxiv.org/abs/2608.01189"}, {"id": "gh:hitdisc/made", "kind": "repo", "url": "https://github.com/HITDiSC/MADE"}], "authors": ["Yicheng Liu", "Bolin Zhang", "Weiran Liu", "Yakun Zhang", "Yangqin Jiang", "Zhiying Tu", "Dianhui Chu"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.01189", "source_url": "https://arxiv.org/abs/2608.01189"}, "categories": ["Agents", "Agents & Tool Use", "General AI"], "description": {"en": "A benchmark of 122 real-world AI models in which agents must deploy each model as a ready-to-call API. Standardized test cases check the result, and agents are compared by deployment success rate."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.01189", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.01189", "title": "M2ABench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Harbin Institute of Technology", "University of Hong Kong"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2608.01189", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2608.01189", "source_url": "https://openalex.org/W7172446615"}, "key": "claire-radar:2608.01189", "modality": null, "name": "M2ABench", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"admission_policy_version": "2026-09-09.1", "confidence": "Medium", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-09-25T04:40:22Z", "data_status": "primary-source-indexed", "display_eligible": "true", "origin_record_id": "bm_m2abench_0419be66", "origin_source": "arxiv", "recognition_confidence": "0.75", "record_sha256": "8c06198cc38eaa70b4fe7db46d2fb1066dca36cf4a8d18c2c93c432f1d5a44d3", "relation": "introduces", "review_model": "claude-code-default", "review_state": "ai-reviewed", "reviewed_at": "2026-09-24T16:38:13.678431Z", "source_url": "https://arxiv.org/abs/2608.01189"}, "publisher": null, "released": "2026-08-02", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.01189", "source_url": "https://arxiv.org/abs/2608.01189"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-01189", "source": "claire_radar", "source_benchmark_id": "2608.01189", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Agents & Tool Use", "attentionForecast": {"confidence": "Low", "horizon": "7d", "reason": "The benchmark is secondary to a method paper, so how many other teams adopt it independently is uncertain.", "score": 25}, "capabilities": [], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-24T16:38:13.678431Z", "inputHash": "9145e96d2a4e41a4c06eca889eab1f39dd4642ea1d5154b6541f37fb347f7a09", "model": "claude-code-default"}, "curation": {"admissionPolicyVersion": "2026-09-09.1", "canonicalNameEvidence": "We further introduce M2ABench, a benchmark for the task of transforming Models to ready-to-call APIs.", "canonicalNameSource": "abstract", "decisionReason": "It is a formally named benchmark with a fixed set of 122 models and standardized test cases, and its code and dataset are publicly available. However, the paper's main contribution is the MADE method.", "evaluationEvidence": {"reuse": "Our code and dataset are publicly available", "scoring": "M2ABench comprises 122 real-world models with standardized test cases for evaluation.", "task": "M2ABench, a benchmark for the task of transforming Models to ready-to-call APIs"}, "model": "claude-code-default", "reviewedAt": "2026-09-24T16:38:13.678431Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "A benchmark of 122 real-world AI models in which agents must deploy each model as a ready-to-call API. Standardized test cases check the result, and agents are compared by deployment success rate.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["exact named benchmark artifact released in abstract", "evaluation protocol evidence", "public artifact URL"], "snippet": "We further introduce M2ABench, a benchmark for the task of transforming Models to ready-to-call APIs."}, "familyId": "bmf_97bafe8171c0", "firstSeenAt": "2026-08-19", "heat": null, "id": "bm_m2abench_0419be66", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/HITDiSC/MADE", "data": null, "paper": "https://arxiv.org/abs/2608.01189", "pdf": "https://arxiv.org/pdf/2608.01189", "project": null, "report": "https://arxiv.org/abs/2608.01189"}, "metrics": [], "motivation": "LLM-based agents now have strong general capabilities.", "name": "M2ABench", "oneLine": "A benchmark of 122 real-world AI models in which agents must deploy each model as a ready-to-call API. Standardized test cases check the result, and agents are compared by deployment success rate.", "paperTitle": "MADE: Belief-Driven Dual-Agent Coordination for Autonomous Model Deployment", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2608.01189", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-08-19T11:05:13.395391Z"}, "publications": [], "readiness": "Runnable", "recognitionConfidence": 0.75, "relation": "introduces", "releasedAt": "2026-08-02", "reviewLifecycle": {"state": "pending"}, "source": {"authors": ["Yicheng Liu", "Bolin Zhang", "Weiran Liu", "Yakun Zhang", "Yangqin Jiang", "Zhiying Tu", "Dianhui Chu"], "categories": ["cs.SE"], "id": "2608.01189", "title": "MADE: Belief-Driven Dual-Agent Coordination for Autonomous Model Deployment", "type": "arxiv", "url": "https://arxiv.org/abs/2608.01189"}, "sourceUpdatedAt": "2026-08-02T12:14:43Z", "topics": ["Agents"], "venueAttempts": [], "whyItMatters": "It targets automated model deployment, a long-horizon, multi-stage agent task that has received little study, and it lets agent systems be compared on success rate against baselines such as SWE-agent and OpenHands."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
