{"generated_at":"2026-10-07T16:10:57Z","site_name":"Decision Models","site_tagline":"决策模型资料库：模型、评测与排行榜","counts":{"models":193,"models_with_scores":185,"models_open_weights":82,"models_multimodal":28,"models_finetunable":59,"benchmarks":246,"benchmarks_s1mb":137,"benchmarks_di":53,"benchmarks_vision":11,"benchmarks_extra":45,"scores":17791,"news":14,"glossary":36,"papers":1192,"papers_topics":20},"domain":"decision.host","home_url":"https://decision.host/","suites":[{"id":"s1mb","label":"S1MB","full":"S1MB · System One Mosaic Benchmark","desc":"英文纯文本决策基准，137 个基准 / 106 个数据子集，三类原语分别计分。","url":"https://huggingface.co/spaces/hotchpotch/S1MB-leaderboard","metric":"Task Avg（基线校正，0–100）","direction":"higher","modalities":["text"]},{"id":"di","label":"Decision Index","full":"Decision Index 0.3（文本主榜）","desc":"跨厂商决策模型综合榜，37 个计分基准分 5 个领域，Full Score = 0.2×public + 0.5×same_skills + 0.3×new_domains。","url":"https://huggingface.co/spaces/multimodalart/jev-decision-index","metric":"Full Score（0–100）","direction":"higher","modalities":["text"]},{"id":"di_vision","label":"Decision Index · Vision","full":"Decision Index — Vision（多模态子榜）","desc":"带图像输入的决策模型子榜，11 个基准，public 与 private 各占一半权重。","url":"https://huggingface.co/spaces/multimodalart/jev-decision-index","metric":"Vision Full Score（0–100）","direction":"higher","modalities":["text","image"]}],"upstream":{"s1mb":{"label":"S1MB","counts":{"benchmarks":137,"models":98,"scores":13295},"generated":"2026-10-07T03:40:04.364Z","revision":"0a3f668f51a405795651aefc82be4fcbf8a1af3e","url":"https://github.com/hotchpotch/S1MB","leaderboard":"https://huggingface.co/spaces/hotchpotch/S1MB-leaderboard"},"di":{"label":"Decision Index","generated":"2026-10-06T17:39:21Z","requests":110201,"panel":37,"url":"https://huggingface.co/spaces/multimodalart/jev-decision-index"},"openrouter":{"label":"OpenRouter decisions","count":14,"url":"https://openrouter.ai/rankings/decisions"},"som":{"label":"systemonemodels.org","counts":{"models":14,"guides":13,"comparisons":6,"integrations":7,"use_cases":18},"url":"https://systemonemodels.org/"},"arxiv":{"label":"arXiv","counts":{"papers":1192,"raw_entries":1845,"topics":20,"newest":"2026-10-06"},"url":"https://arxiv.org/","topics":20}},"filter_options":{"vendors":["社区","Jared Palmer","vLLM Semantic Router","Intern Large Model","Zefan Cai","Mapika","Fastino Labs","flymy-ai","mstrasser","AutoTrust AI Lab","Frontier Infra","HopitAI","Together AI","hotchpotch","AlexWortega","Perplexity","Cloudflare","Juspay Technologies","EldanRing","Sthānika AI","Hanno-Labs","Convai Innovations","Respan","AnkitAI","apus-ailab","dhtocks","Fastino","TypeSafe AI","Torchcast AI","Pixilab.ai","Surogate","社区（alibiserikbay）","Bespoke Labs","Doccy Health","TokenRhythm","Interfaze","wayfind","DanielMcSheehy","FLock.io","OmniJev","Cortex Agent LLC","wfzyx","knowledgator","Contrastive-LM","Mogan AI","Danger Labs","Supersonic Labs","Manav2op","OpenAI","Liquid AI","Inception","Upstage","meraGPT","openjev（社区）","denis-pplx","zeredy879","未标注","LocalLLaMA","abhishek085","Alibaba-NLP","altslate","AndeyTait","Cactus-Compute","caiovicentino1","crh225","evalengine","facebook","franckverrot","FrontiersMind","GestaltLabs","heman10x","iapp","isHeSatoshi","jsaurabh","kshetrajna12","LibertAIDAI","mghafiri","mixedbread-ai","mohit67890","MoritzLaurer","oraculumai","PIXELZX","pngwn","Qwen","StandardThinking","togethercomputer","zeroentropy"],"categories":["commercial","dataset","open","tiny"],"licenses":["未标注","Apache-2.0","闭源","即将开源","许可证未定稿","Apache-2.0（权重）/ MIT（代码）","未公开"],"license_kinds":["other","permissive","unknown"],"benchmark_suites":["s1mb","di","di_vision","通用多模态","Agent 决策"],"paper_topics":["决策模型","System One 模型","System 1 双过程","Jev / TypeSafe","类型化决策","校准决策","决策头 / 读出","单次前向 / prefill 读出","LLM 路由","奖励模型与校准","护栏与安全分类","蒸馏到小模型","多模态决策","结构化输出","工具调用评测","logit 读出与分类","弃权与拒答","LLM 校准","严格评分规则","面向决策的重排 / 推荐"],"paper_genres":["其他","基准与评测","安全与治理","应用与落地","方法与训练","系统与工程"],"benchmark_groups":["公开 NLP 数据集","Open-Jev 运维与控制","B 通用多模态","检索相关性（NanoBEIR / 合成）","C Agent 与工具调用","知识与推理","多模态决策（图像输入）","语言理解","艺术与人类品味","检索与分类","S1MB 泛化（合成）","工具与自动化","knowledge","Laya 数据集","tools","games","retrieval"],"decision_types":["choice","noul","score","mixed"]},"source_index":{"s1mb.viewer_summary":{"url":"https://huggingface.co/datasets/hotchpotch/s1mb-result/resolve/main/viewer-summary.json","fetched_at":1791373789.9667451,"bytes":7095755},"s1mb.category_english":{"url":"https://raw.githubusercontent.com/hotchpotch/S1MB/main/evaluator/data/categories/english-v1.json","fetched_at":1791373790.561002,"bytes":6165},"s1mb.readme":{"url":"https://raw.githubusercontent.com/hotchpotch/S1MB/main/README.md","fetched_at":1791373791.0478458,"bytes":6298},"s1mb.benchmark_scope":{"url":"https://raw.githubusercontent.com/hotchpotch/S1MB/main/docs/benchmark_scope.md","fetched_at":1791373791.542038,"bytes":5232},"s1mb.adapters":{"url":"https://raw.githubusercontent.com/hotchpotch/S1MB/main/docs/adapters.md","fetched_at":1791373792.0166368,"bytes":5523},"s1mb.third_party":{"url":"https://raw.githubusercontent.com/hotchpotch/S1MB/main/THIRD_PARTY_NOTICES.md","fetched_at":1791373793.6581728,"bytes":864},"s1mb.dataset_readme":{"url":"https://huggingface.co/datasets/hotchpotch/s1mb-dataset/raw/main/README.md","fetched_at":1791373794.3754199,"bytes":42830},"s1mb.blog_article":{"url":"https://huggingface.co/blog/hotchpotch/system-one-mosaic-benchmark","fetched_at":1791373797.0161102,"bytes":245078},"s1mb.bekko_article":{"url":"https://huggingface.co/blog/hotchpotch/bekko-system-one-v0-release","fetched_at":1791373801.996999,"bytes":279672},"di.index":{"url":"https://huggingface.co/spaces/multimodalart/jev-decision-index/raw/main/data/index.json","fetched_at":1791373826.877286,"bytes":1503478},"di.methodology":{"url":"https://huggingface.co/spaces/multimodalart/jev-decision-index/raw/main/data/methodology.json","fetched_at":1791373830.5723038,"bytes":293389},"di.v03":{"url":"https://huggingface.co/spaces/multimodalart/jev-decision-index/raw/main/data/v03.json","fetched_at":1791373831.410149,"bytes":31981},"di.vision":{"url":"https://huggingface.co/spaces/multimodalart/jev-decision-index/raw/main/data/vision.json","fetched_at":1791373832.12387,"bytes":28893},"di.own":{"url":"https://huggingface.co/spaces/multimodalart/jev-decision-index/raw/main/data/own.json","fetched_at":1791373832.856823,"bytes":26282},"di.news":{"url":"https://huggingface.co/spaces/multimodalart/jev-decision-index/raw/main/news.html","fetched_at":1791373833.86355,"bytes":88819},"di.readme":{"url":"https://huggingface.co/spaces/multimodalart/jev-decision-index/raw/main/README.md","fetched_at":1791373834.5792441,"bytes":6675},"or.decision_models":{"url":"https://openrouter.ai/api/v1/models?output_modalities=decisions","fetched_at":1791373835.2224782,"bytes":12774},"or.all_models":{"url":"https://openrouter.ai/api/v1/models","fetched_at":1791373844.543722,"bytes":774404},"or.jev_docs":{"url":"https://openrouter.ai/docs/guides/community/jev","fetched_at":1791373856.10565,"bytes":534026},"som.home":{"url":"https://systemonemodels.org/","fetched_at":1791373856.846554,"bytes":170092},"som.models":{"url":"https://systemonemodels.org/models","fetched_at":1791373857.983867,"bytes":75514},"som.use_cases":{"url":"https://systemonemodels.org/use-cases","fetched_at":1791373859.0517719,"bytes":82428},"som.recipes":{"url":"https://systemonemodels.org/recipes","fetched_at":1791373860.491201,"bytes":48679},"som.glossary":{"url":"https://systemonemodels.org/glossary","fetched_at":1791373861.7080262,"bytes":60670},"som.llms_txt":{"url":"https://systemonemodels.org/llms.txt","fetched_at":1791373862.4533272,"bytes":21934},"cf.clef_blog":{"url":"https://blog.cloudflare.com/clef-decision-models/","fetched_at":1791373864.3171608,"bytes":315642},"or.decisions_api_ref":{"url":"https://openrouter.ai/docs/api/api-reference/alphadecisions/submit-a-decisions-request.md","fetched_at":1791373864.975539,"bytes":56554},"oai.decisions":{"url":"https://developers.openai.com/api/docs/guides/decisions.md","fetched_at":1791373865.6482902,"bytes":29581},"pplx.decisions":{"url":"https://docs.perplexity.ai/docs/decisions/quickstart.md","fetched_at":1791373866.734153,"bytes":19098},"awesome.system_one":{"url":"https://raw.githubusercontent.com/pozapas/awesome-system-one-models/main/README.md","fetched_at":1791373870.013781,"bytes":151063},"arxiv.decision-model":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22decision+model%22&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791380905.5689511,"bytes":293161},"arxiv.system-one":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22System+One+model%22&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791380911.6884532,"bytes":67544},"arxiv.system-1":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22System+1%22+AND+cat%3Acs.CL&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791380922.223,"bytes":252650},"arxiv.jev":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22Jev%22&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791380931.4452648,"bytes":209418},"arxiv.typed-decision":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22typed+decision%22&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791380937.3900669,"bytes":104289},"arxiv.calibrated-decision":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22calibrated+decision%22&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791380944.243293,"bytes":139570},"arxiv.decision-head":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22decision+head%22&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791380950.2665422,"bytes":61408},"arxiv.prefill":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22prefill%22+AND+all%3A%22decision%22&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791380959.366478,"bytes":197419},"arxiv.llm-routing":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22LLM+routing%22&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791380964.056363,"bytes":291145},"arxiv.reward-calibration":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22reward+model%22+AND+all%3A%22calibration%22&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791380974.630433,"bytes":296942},"arxiv.guardrail":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22guardrail%22+AND+all%3A%22classifier%22&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791380984.05056,"bytes":232778},"arxiv.distillation":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22distillation%22+AND+all%3A%22classifier%22&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791380994.0846221,"bytes":302330},"arxiv.multimodal-decision":{"url":"https://export.arxiv.org/api/query?search_query=abs%3A%22multimodal%22+AND+abs%3A%22decision%22+AND+cat%3Acs.CL&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791381008.860369,"bytes":315908},"arxiv.structured-output":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22structured+output%22+AND+cat%3Acs.CL&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791381022.6501918,"bytes":298311},"arxiv.function-calling":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22function+calling%22+AND+all%3A%22evaluation%22&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791381034.071281,"bytes":316989},"arxiv.logit-classification":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22logit%22+AND+all%3A%22classification%22+AND+cat%3Acs.CL&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791381039.692254,"bytes":137732},"arxiv.abstention":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22abstention%22+AND+cat%3Acs.CL&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791381052.012084,"bytes":300544},"arxiv.calibration-llm":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22calibration%22+AND+cat%3Acs.CL&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791381061.9681711,"bytes":300266},"arxiv.proper-scoring":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22proper+scoring+rule%22&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791381072.522885,"bytes":276430},"arxiv.decision-oriented":{"url":"https://export.arxiv.org/api/query?search_query=all%3A%22decision-oriented%22&max_results=120&sortBy=submittedDate&sortOrder=descending","fetched_at":1791381077.238873,"bytes":187084}},"data_notes":["S1MB 与 Decision Index 的分数口径不同，不可互相换算。","「未出现在某榜」不等于 0 分；本站在详情页区分「未参评」与「得分低」。","S1MB 的 Task Avg 是基线校正指数，不是准确率。","Decision Index 的 Full Score 含未公开的私有集，纯公开口径请看 public 列。","S1MB 必须看覆盖率：未跑满 137 个基准的模型不参与排名。"],"disclaimer":"本站聚合自 S1MB、Decision Index、OpenRouter、systemonemodels.org 等公开来源，所有分数与链接均标注出处；模型名称与商标归各自所有者。"}