{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2605.27820", "kind": "paper", "url": "https://arxiv.org/abs/2605.27820"}], "authors": ["Yunqi Liu", "Tong Niu", "Zitong Wang", "Zhenlong Dai", "Yuqi Qing", "Weiqiang Wang", "Jian Liu"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2605.27820", "source_url": "https://arxiv.org/abs/2605.27820"}, "categories": ["Agents & Tool Use", "General AI", "Multimodal"], "description": {"en": "EgoBench is an interactive multimodal benchmark for tool-using agents. It comprises 1,045 egocentric-video-grounded tasks across four daily scenarios, with a user-agent-tool interactive environment. It assesses multimodal perception, tool invocation with multi-hop reasoning, and dynamic user interaction."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2605.27820", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2605.27820", "title": "EgoBench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2605.27820", "modality": null, "name": "EgoBench", "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_egobench_d0f8580c", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "75581f3adf59d26971cb36c19c390cef8741be803740e3aa84abec88562ac3ad", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2605.27820"}, "publisher": null, "released": "2026-05-27", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2605.27820", "source_url": "https://arxiv.org/abs/2605.27820"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2605-27820", "source": "claire_radar", "source_benchmark_id": "2605.27820", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Agents & Tool Use", "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/2605.27820", "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", "lastSuccessfulDate": "2026-09-30", "state": "stale"}}, "status": "stale"}, "capabilities": [], "confidence": "High", "construction": "Interactive Environment", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "51ad36a0df77af5d2ff7bfe561f6ac08591c213b89d6f85fee1d5df9e6633fd5", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "EgoBench is an interactive multimodal benchmark for tool-using agents. It comprises 1,045 egocentric-video-grounded tasks across four daily scenarios, with a user-agent-tool interactive environment. It assesses multimodal perception, tool invocation with multi-hop reasoning, and dynamic user interaction.", "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": "To bridge this gap, we introduce EgoBench, the first interactive multimodal benchmark for tool-using agents."}, "familyId": "bmf_31eeb47a8e16", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_egobench_d0f8580c", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2605.27820", "paper": "https://arxiv.org/abs/2605.27820", "pdf": "https://arxiv.org/pdf/2605.27820", "project": null, "report": "https://arxiv.org/abs/2605.27820"}, "metrics": [], "motivation": "As AI agents increasingly operate in open, real-world environments, they require a deep synergy of multimodal perception, tool invocation with multi-hop reasoning, and dynamic interaction with users.", "name": "EgoBench", "oneLine": "EgoBench is an interactive multimodal benchmark for tool-using agents. It comprises 1,045 egocentric-video-grounded tasks across four daily scenarios, with a user-agent-tool interactive environment. It assesses multimodal perception, tool invocation with multi-hop reasoning, and dynamic user interaction.", "paperTitle": "EgoBench: An Interactive Egocentric Multimodal Benchmark for Tool-Using Agents", "primaryDomain": "General AI", "publication": {"evidence": "Accepted to NeurIPS 2026. Camera-ready version", "evidenceLevel": "author-claim", "evidenceUrl": "https://arxiv.org/abs/2605.27820", "source": "arxiv-comments", "status": "acceptance_claimed", "venue": "NeurIPS 2026", "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "ranking": {}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-05-27", "source": {"authors": ["Yunqi Liu", "Tong Niu", "Zitong Wang", "Zhenlong Dai", "Yuqi Qing", "Weiqiang Wang", "Jian Liu"], "categories": ["cs.AI"], "id": "2605.27820", "title": "EgoBench: An Interactive Egocentric Multimodal Benchmark for Tool-Using Agents", "type": "arxiv", "url": "https://arxiv.org/abs/2605.27820"}, "sourceUpdatedAt": "2026-05-27T01:28:15Z", "topics": ["Multimodal"], "venueAttempts": [{"decisionRaw": "Accepted to NeurIPS 2026. Camera-ready version", "evidence": [{"level": "author-claim", "observedAt": "2026-10-14T01:46:58.058199Z", "rawValue": "Accepted to NeurIPS 2026. Camera-ready version", "sourceType": "arxiv-comments", "sourceUrl": "https://arxiv.org/abs/2605.27820"}], "reviewStatus": "accepted", "venueName": "NeurIPS 2026"}], "whyItMatters": "AI agents in open environments need joint multimodal and tool-use capabilities. EgoBench provides a standardized interactive environment and deterministic joint validation to objectively measure these skills, addressing a lack of comparable benchmarks for dynamic tool-using agents."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
