[{"data":1,"prerenderedAt":82},["ShallowReactive",2],{"tool-flageval":3},{"tool":4,"relatedTools":44},{"id":5,"name":6,"slug":7,"tagline":8,"description":9,"websiteUrl":10,"logoText":11,"pricing":12,"editorScore":13,"clickCount":14,"featured":15,"latest":15,"updatedAt":16,"category":17,"tags":26,"overview":28,"highlights":29,"useCases":38},"dbc5979bc608bcdde563fd7a121cbe66","FlagEval","flageval","智源研究院推出的FlagEval（天秤）大模型评测平台","FlagEval（天秤）是北京智源人工智能研究院（BAAI）推出的科学、公正、开放的大模型评测体系及开放平台，为研究人员提供全面评估基础模型及训练算法性能的工具和方法。FlagEval采用“能力-任务-指标”三维评测框架，从多个维度对大模型的认知能力进行评估，涵盖对话、问答、情感分析等多种应用场景，提供超过22个数据集和8万道评测题目。","https:\u002F\u002Fflageval.baai.ac.cn\u002F#\u002Ftrending","FLA","官网可查看",98,9,false,"2026-04-15T04:26:34+08:00",{"id":18,"name":19,"slug":20,"description":21,"iconLabel":22,"accentColor":23,"toolCount":24,"sortOrder":25},"d6d303a9f41677e5f1a5f9b12c520225","智能体与自动化","agents","适合搭建业务工作流、插件编排和多步骤自动化，用来把 AI 接进真实流程。","AUTO","#6C7A2D",216,7,[27],"AI模型评测","FlagEval（天秤）是北京智源人工智能研究院（BAAI）推出的科学、公正、开放的大模型评测体系及开放平台，为研究人员提供全面评估基础模型及训练算法性能的工具和方法。FlagEval采用“能力-任务-指标”三维评测框架，从多个维度对大模型的认知能力进行评估，涵盖对话、问答、情感分析等多种应用场景，提供超过22个数据集和8万道评测题目。平台支持多模态模型评测，覆盖文本、图像、视频等多种数据类型，兼容多种AI框架和硬件架构。FlagEval提供自动化评测机制，支持主观与客观评测的全自动流水线，帮助研究人员高效、准确地了解模型性能，推动大模型技术的发展。 多维度评测框架 ：采用“能力-任务-指标”三维评测框架，从多个维度全面评估大模型的认知能力，涵盖对话、问答、情感分析等多种应用场景。",[30,31,32,33,34,35,36,37],"多维度评测框架","丰富的评测数据集","多模态支持","自动化评测机制","广泛的模型覆盖","排行榜与结果展示","社区参与与持续更新","注册与登录",[39,40,41,42,43,33],"对话","问答","情感分析等多种应用场景","图像","视频等多种数据类型",[45,58,69],{"id":46,"name":47,"slug":48,"tagline":49,"description":50,"websiteUrl":51,"logoText":52,"pricing":12,"editorScore":13,"clickCount":53,"featured":54,"latest":54,"updatedAt":16,"category":55,"tags":56},"a638bfd4cc081d0294d7d12975a727dd","GPT-4","gpt-4","OpenAI旗下最新的GPT-4模型","GPT-4（Generative Pre-trained Transformer）是 OpenAI 开发的自然语言处理模型 GPT 家族中的第四个版本，该模型依靠强大的神经网络来理解和生成类似人类的语言。","https:\u002F\u002Fcn.bing.com\u002Fsearch?q=GPT-4&ensearch=1","GPT",2356,true,{"id":18,"name":19,"slug":20,"description":21,"iconLabel":22,"accentColor":23,"toolCount":24,"sortOrder":25},[57],"AI训练模型",{"id":59,"name":60,"slug":61,"tagline":62,"description":63,"websiteUrl":64,"logoText":65,"pricing":12,"editorScore":13,"clickCount":66,"featured":54,"latest":54,"updatedAt":16,"category":67,"tags":68},"a23e89d9e4203bd1c683c41e837f370e","LMArena","lmarena","AI模型评估平台","LMArena是加州大学伯克利分校推出的创新AI模型评估平台，基于让用户对不同AI模型的回答进行匿名投票，衡量模型的表现。用户输入问题后，平台提供两个模型的回答，用户根据偏好选择更优答案，投票结果直接塑造公共排行榜。","https:\u002F\u002Flmarena.ai\u002F?utm_source=ai-bot.cn","LMA",681,{"id":18,"name":19,"slug":20,"description":21,"iconLabel":22,"accentColor":23,"toolCount":24,"sortOrder":25},[27],{"id":70,"name":71,"slug":72,"tagline":73,"description":74,"websiteUrl":75,"logoText":76,"pricing":77,"editorScore":13,"clickCount":78,"featured":54,"latest":54,"updatedAt":16,"category":79,"tags":80},"32bb7d109995125694dc4a175a840117","Coze","coze","海量AI智能体免费用，一键复制同款","Coze是字节跳动推出的AI办公助手，能通过AI技术简化工作流程，提升工作效率。通过Agent Skills技能系统和Agent Plan长期计划功能，实现从单一工具到智能工作伙伴的转变。","https:\u002F\u002Fwww.coze.cn\u002Foverview?utm_medium=daohang&utm_source=aibot&utm_term=49156769m","COZ","免费",316,{"id":18,"name":19,"slug":20,"description":21,"iconLabel":22,"accentColor":23,"toolCount":24,"sortOrder":25},[81],"AI开发平台",1776262185614]