{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.29387", "kind": "paper", "url": "https://arxiv.org/abs/2608.29387"}, {"id": "gh:maps-research/evogenui-bench", "kind": "repo", "url": "https://github.com/MAPS-research/EvoGenUI-Bench"}], "authors": ["Yue Peng", "Lanke Xia", "Zihan Wang", "Jiahao Ye", "Ke Ning", "Hongyi Wen"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2608.29387", "source_url": "https://arxiv.org/abs/2608.29387"}, "categories": ["General AI", "Language & Knowledge", "cs.AI"], "description": {"en": "Evaluates large language models as multi-turn generative UI assistants on 150 five-turn tasks with browser-executed artifacts, using turn-level, episode-level, and Adjacent Pass Retention metrics."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.29387", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.29387", "title": "EvoGenUI-Bench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2608.29387", "modality": null, "name": "EvoGenUI-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": "true", "origin_record_id": "bm_evogenui-bench_3b239061", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "0ddf7bee0b119753cd96f7b4ecc5fafc2073919b0b9613a005d33cdd70f5b6db", "relation": "introduces", "review_model": "deepseek-v4-pro", "review_state": "ai-reviewed", "reviewed_at": "2026-09-02T05:23:11.715633Z", "source_url": "https://arxiv.org/abs/2608.29387"}, "publisher": null, "released": "2026-08-29", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.29387", "source_url": "https://arxiv.org/abs/2608.29387"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-29387", "source": "claire_radar", "source_benchmark_id": "2608.29387", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Language & Knowledge", "attention": {"asOf": "2026-09-02", "attemptedAt": "2026-09-02T05:12:51.621571Z", "githubRepo": "https://github.com/MAPS-research/EvoGenUI-Bench", "githubScope": "benchmark_repo", "githubStars": 0, "hfDailySubmittedAt": "2026-09-01T00:00:00.000Z", "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 4, "hfPaperUrl": "https://huggingface.co/papers/2608.29387", "observedAt": "2026-09-02T05:12:51.621571Z", "signalStatus": {"githubStars": {"observedAt": "2026-09-02T05:12:51.621571Z", "state": "fresh"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"observedAt": "2026-09-02T05:12:51.621571Z", "state": "fresh"}}, "status": "fresh"}, "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "The benchmark targets a rapidly emerging area of generative UI and provides a complete evaluation framework with public code and leaderboard, which typically attracts both developers and researchers.", "score": 74}, "capabilities": [], "confidence": "High", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-02T05:23:11.715633Z", "inputHash": "41d82ef36eb138f0ecd4090559226ad1d5c9936b8ac9bca223af67dd343366e4", "model": "deepseek-v4-pro"}, "curation": {"canonicalNameEvidence": "We introduce EvoGenUI-Bench, a benchmark for multi-turn interface maintenance comprising 150 five-turn tasks and 750 turns across three scenarios: information presentation, executable interaction, and tool-grounded external state.", "canonicalNameSource": "abstract", "decisionReason": "Formally named benchmark with a stable scoring protocol, public code repository, and explicit leaderboard infrastructure.", "model": "deepseek-v4-pro", "reviewedAt": "2026-09-02T05:23:11.715633Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "Evaluates large language models as multi-turn generative UI assistants on 150 five-turn tasks with browser-executed artifacts, using turn-level, episode-level, and Adjacent Pass Retention metrics.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "score_submission", "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence", "evaluation protocol evidence"], "snippet": "We introduce EvoGenUI-Bench, a benchmark for multi-turn interface maintenance comprising 150 five-turn tasks and 750 turns across three scenarios: information presentation, executable interaction, and tool-grounded external state."}, "familyId": "bmf_119fd5f72bb5", "firstSeenAt": "2026-09-01", "heat": null, "id": "bm_evogenui-bench_3b239061", "indexedAt": "2026-09-01T06:28:29.895112Z", "industrySectors": [], "links": {"code": "https://github.com/MAPS-research/EvoGenUI-Bench", "data": null, "hfPaper": "https://huggingface.co/papers/2608.29387", "paper": "https://arxiv.org/abs/2608.29387", "pdf": "https://arxiv.org/pdf/2608.29387", "project": null, "report": "https://arxiv.org/abs/2608.29387"}, "metrics": [], "motivation": "Large language models can generate interactive web interfaces, but reliable generative UI requires maintaining an executable artifact as user requests evolve.", "name": "EvoGenUI-Bench", "oneLine": "Evaluates large language models as multi-turn generative UI assistants on 150 five-turn tasks with browser-executed artifacts, using turn-level, episode-level, and Adjacent Pass Retention metrics.", "paperTitle": "EvoGenUI-Bench: Evaluating LLMs as Multi-Turn Generative UI Assistants", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2608.29387", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-09-02T00:50:41.903960Z"}, "publications": [], "publishers": [{"name": "MAPS-research", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://github.com/MAPS-research/EvoGenUI-Bench"}], "ranking": {"30d": {"confidence": "High", "coverage": 0.85, "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": 30}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": 0.0, "value": 0}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.285264874482679, "value": 4}}, "confidence": "High", "coverage": 0.85, "rank": 190, "score": 16}, "method": "attention-ranking-v13", "rank": 190, "score": 16}, "90d": {"confidence": "Medium", "coverage": 0.7, "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": 0.0, "value": 0}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": 0.285264874482679, "value": 4}}, "confidence": "Medium", "coverage": 0.7, "rank": 404, "score": 19}, "method": "attention-ranking-v13", "rank": 404, "score": 19}}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-08-29", "source": {"authors": ["Yue Peng", "Lanke Xia", "Zihan Wang", "Jiahao Ye", "Ke Ning", "Hongyi Wen"], "categories": ["cs.AI", "cs.SE"], "id": "2608.29387", "title": "EvoGenUI-Bench: Evaluating LLMs as Multi-Turn Generative UI Assistants", "type": "arxiv", "url": "https://arxiv.org/abs/2608.29387"}, "sourceUpdatedAt": "2026-08-29T17:54:02Z", "topics": ["cs.AI"], "venueAttempts": [], "whyItMatters": "Shifts generative UI evaluation from isolated outputs to multi-turn consistency, with a comprehensive public harness and leaderboard that enables detailed model comparisons."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
