{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2608.28062", "kind": "paper", "url": "https://arxiv.org/abs/2608.28062"}], "categories": ["General AI", "Information retrieval", "Multimodal", "Planning"], "description": {"en": "VisTarget-Bench is a 150-task human-verified benchmark pairing each multimodal search question with a held-out target image to separate image-retrieval failures from visual-perception failures."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2608.28062", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2608.28062", "title": "VisTarget-Bench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2608.28062", "modality": null, "name": "VisTarget-Bench", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "Medium", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "false", "origin_record_id": "bm_vistarget-bench_aaa3c19c", "origin_source": "arxiv", "recognition_confidence": "0.55", "record_sha256": "00b6f6a790b5738f86d4f809efc475211abba9bf6bd07b4679fe1922a111f1d5", "relation": "evaluates_only", "review_model": "deepseek-v4-pro", "review_state": "ai-name-audit-deferred", "reviewed_at": "2026-09-03T09:04:52.212719Z", "source_url": "https://arxiv.org/abs/2608.28062"}, "publisher": null, "released": "2026-08-28", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2608.28062", "source_url": "https://arxiv.org/abs/2608.28062"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2608-28062", "source": "claire_radar", "source_benchmark_id": "2608.28062", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Multimodal", "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/2608.28062", "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"}, "attentionForecast": {"confidence": "Low", "horizon": "7d", "reason": "The benchmark targets a narrow but currently active multimodal agent area, yet the abstract lacks a public release artifact or adoption signal, limiting forecast confidence.", "score": 42}, "capabilities": ["Planning", "Information retrieval"], "confidence": "Medium", "construction": "Transform Existing", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-03T09:04:52.212719Z", "inputHash": "20447697dcbd060d0214fb5c674e3610afe9592b6e262b5dca2de633ff17257b", "model": "deepseek-v4-pro"}, "curation": {"canonicalNameEvidence": "We also introduce VisTarget-Bench, a 150-task human-verified benchmark that pairs each question with a held-out target image, distinguishing image-retrieval failures from visual-perception failures.", "canonicalNameSource": "abstract", "decisionReason": "The benchmark is formally announced in the abstract with a defined evaluation object and scoring purpose, though no public reuse link is provided in the evidence.", "model": "deepseek-v4-pro", "reviewedAt": "2026-09-03T09:04:52.212719Z", "state": "ai-name-audit-deferred"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "VisTarget-Bench is a 150-task human-verified benchmark pairing each multimodal search question with a held-out target image to separate image-retrieval failures from visual-perception failures.", "displayEligible": false, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["exact named benchmark artifact released in abstract", "evaluation protocol evidence", "named entity is a model, method, framework, or dataset used in evaluation"], "snippet": "During post-training, our Failure-Aware GSPO (FA-GSPO) recovers salvageable abnormal rollouts and filters invalid ones to improve bounded multimodal planning and search.We also introduce VisTarget-Bench, a 150-task human-verified benchmark that pairs each question with a held-out target image, distinguishing image-retrieval failures from visual-perception failures."}, "familyId": "bmf_a8b611d92a57", "firstSeenAt": "2026-08-31", "heat": null, "id": "bm_vistarget-bench_aaa3c19c", "indexedAt": "2026-08-31T01:03:30.163531Z", "industrySectors": [], "links": {"code": null, "data": null, "paper": "https://arxiv.org/abs/2608.28062", "pdf": "https://arxiv.org/pdf/2608.28062", "project": null, "report": "https://arxiv.org/abs/2608.28062"}, "metrics": [], "motivation": "Multimodal search agents extend parametric knowledge with newly emerging and long-tail evidence from the open web.", "name": "VisTarget-Bench", "oneLine": "VisTarget-Bench is a 150-task human-verified benchmark pairing each multimodal search question with a held-out target image to separate image-retrieval failures from visual-perception failures.", "paperTitle": "WeAgent-MMSearch: Native Text-Vision Interaction for Multimodal Search Agents", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2608.28062", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-08-31T01:03:30.163531Z"}, "publications": [], "ranking": {"90d": {"confidence": "Low", "coverage": 0.0, "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": null, "value": null}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "rank": null, "score": null}, "method": "attention-ranking-v13", "rank": null, "score": null}}, "readiness": "Paper only", "recognitionConfidence": 0.55, "relation": "evaluates_only", "releasedAt": "2026-08-28", "source": {"authors": ["Zongkai Liu", "Hui Zhang", "Liqiang Niu", "Zhen Cao", "Han Li", "Juntao Liu", "Wenchao Chen", "Chengduo Zhao", "Chao Yu", "Fandong Meng"], "categories": ["cs.AI"], "id": "2608.28062", "title": "WeAgent-MMSearch: Native Text-Vision Interaction for Multimodal Search Agents", "type": "arxiv", "url": "https://arxiv.org/abs/2608.28062"}, "sourceUpdatedAt": "2026-08-28T08:28:43Z", "topics": ["Multimodal"], "venueAttempts": [], "whyItMatters": "It isolates whether multimodal agents fail to retrieve the correct image or to perceive it correctly, providing a more granular diagnostic than existing agentic search benchmarks."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
