{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2606.22437", "kind": "paper", "url": "https://arxiv.org/abs/2606.22437"}], "authors": ["Wenzhen Yuan", "Jiacheng Ruan", "Wutao Xiong", "Chengping Zhao", "Ting Liu", "Yuzhuo Fu"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2606.22437", "source_url": "https://arxiv.org/abs/2606.22437"}, "categories": ["General AI", "Multimodal"], "description": {"en": "MMGist is a curated multimodal benchmark covering seven capability dimensions with 7,262 items, derived from 18 existing benchmarks via filtering pipelines. It evaluates vision-language models across dimensions such as Visual Logic and Expert Knowledge."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.22437", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.22437", "title": "MMGist"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2606.22437", "modality": null, "name": "MMGist", "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_mmgist_394485cc", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "928030ca20a79ff3cf05560ff587cec27cc0002ab1198ad023244f179bdfb3b5", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2606.22437"}, "publisher": null, "released": "2026-06-21", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.22437", "source_url": "https://arxiv.org/abs/2606.22437"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-22437", "source": "claire_radar", "source_benchmark_id": "2606.22437", "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": 1, "hfPaperUrl": "https://huggingface.co/papers/2606.22437", "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": "2026-10-04", "state": "stale"}}, "status": "stale"}, "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": "e4221c16868e4b94571867d898969f7c8b4c85170f3ea03cbc2f98f43bb03e61", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "MMGist is a curated multimodal benchmark covering seven capability dimensions with 7,262 items, derived from 18 existing benchmarks via filtering pipelines. It evaluates vision-language models across dimensions such as Visual Logic and Expert Knowledge.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact coined title identity tied to benchmark evidence", "benchmark release stated in one sentence"], "snippet": "To this end, we propose MMGist, a curated benchmark that covers seven capability dimensions and contains 7,262 items."}, "familyId": "bmf_06d36f826ac3", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [], "status": "not-found"}, "heat": null, "id": "bm_mmgist_394485cc", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": null, "data": null, "hfPaper": "https://huggingface.co/papers/2606.22437", "paper": "https://arxiv.org/abs/2606.22437", "pdf": "https://arxiv.org/pdf/2606.22437", "project": null, "report": "https://arxiv.org/abs/2606.22437"}, "metrics": [], "motivation": "We conduct a systematic study of 18 widely used vision-language benchmarks and identify three major issues: 1) many items do not rely on visual cues and therefore fail to effectively measure multimodal understanding; 2) many items are already close to performance saturation for current LVLMs, which limits their discriminative power; 3) a small number of anomalous items affect the reliability of evaluation results.", "name": "MMGist", "oneLine": "MMGist is a curated multimodal benchmark covering seven capability dimensions with 7,262 items, derived from 18 existing benchmarks via filtering pipelines. It evaluates vision-language models across dimensions such as Visual Logic and Expert Knowledge.", "paperTitle": "MMGist: A Comprehensive Multimodal Benchmark for 2027", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2606.22437", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "ranking": {}, "readiness": "Paper only", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-06-21", "source": {"authors": ["Wenzhen Yuan", "Jiacheng Ruan", "Wutao Xiong", "Chengping Zhao", "Ting Liu", "Yuzhuo Fu"], "categories": ["cs.CV", "cs.AI"], "id": "2606.22437", "title": "MMGist: A Comprehensive Multimodal Benchmark for 2027", "type": "arxiv", "url": "https://arxiv.org/abs/2606.22437"}, "sourceUpdatedAt": "2026-06-25T12:10:03Z", "topics": ["Multimodal"], "venueAttempts": [], "whyItMatters": "MMGist addresses the need for reliable and discriminative evaluation of vision-language models, aiming to reduce redundancy and saturation in existing benchmarks while preserving model rankings."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
