[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$fyrABXXY6GFSOgcGPn8USpoMdY6yUydkvW02Fsl3Vk1c":3,"$fWNsfe5LyMk_QM7fXcoufrWVu1aa3Gn4kGdkU0eJxsuI":80},{"id":4,"name":5,"code":6,"version":7,"logo":8,"url":9,"vendorId":10,"vendorName":11,"vendorCode":12,"vendorCountry":13,"modelType":34,"openSource":35,"openLicense":9,"paramSize":36,"contextLength":37,"intro":38,"supportApi":18,"apiUrl":39,"pricingType":34,"freeQuota":40,"score":23,"isRecommend":18,"isHot":18,"click":41,"tags":42,"content":43,"pricings":44,"scores":52},50,"通义千问 Qwen3.8-Max","qwen3.8-max","3.8-Max","\u002Flogos\u002Falibaba.svg",null,8,"阿里云","alibaba",[14,19,22,26,30],{"id":15,"name":16,"color":17,"type":18},5,"支持API","#00B4A0",1,{"id":10,"name":20,"color":21,"type":18},"国产自研","#F56C6C",{"id":23,"name":24,"color":25,"type":18},9,"推理强化","#E65100",{"id":27,"name":28,"color":29,"type":18},10,"长上下文","#3B55D9",{"id":31,"name":32,"color":33,"type":18},11,"多模态理解","#9C27B0",2,0,"2.4T(A95B)",1000000,"阿里 2026 年 8 月发布的旗舰，2.4 万亿参数 + 95B 激活，已接入千问办公等 Agent 产品","https:\u002F\u002Fhelp.aliyun.com\u002Fzh\u002Fmodel-studio","新用户开通后有 90 天有效期内的免费额度，以百炼控制台为准",22,[],"Qwen3.8-Max 于 2026 年 8 月 3 日发布，是 Qwen 家族迄今最强的模型，基于 Qwen3.5 架构，通过稀疏 MoE 架构与混合注意力机制的联合优化实现：总参数扩展至 2.4 万亿、单次推理仅激活 950 亿参数，上下文窗口最长 100 万 token，输入模态覆盖文本、图片与视频，输出为文本，可处理长视频与复杂文档。它在编程、专业办公、科研与长周期任务上较上代全面提升，第三方 Vision Arena 排名全球第二。目前 API 已上线阿里云百炼平台，并同步接入千问办公等 Agent 产品；次日阿里港美股双双大涨，成为估值修复的直接催化。快照版本 qwen3.8-max-0902 已在后续更新中上线。同系列另有 Qwen3.8-27B（开源）、Qwen3.8-Plus 与主打极速推理的 Qwen3.8-Flash（输入 0.8 元 \u002F 输出 2.4 元），本文一并归入本条目。",[45],{"itemName":46,"billingType":18,"inputPrice":47,"outputPrice":48,"price":35,"unit":49,"currency":50,"freeQuota":9,"intro":51},"按量计费",12,36,"元\u002F百万tokens","CNY","缓存命中输入 1.5 元；Batch 模式 6 \u002F 18 元",[53,57,60,64,68,71,75],{"metricId":18,"metricName":54,"metricCode":55,"maxScore":27,"score":56},"中文理解","chinese",9.5,{"metricId":34,"metricName":58,"metricCode":59,"maxScore":27,"score":23},"文本写作","writing",{"metricId":61,"metricName":62,"metricCode":63,"maxScore":27,"score":23},3,"逻辑推理","reasoning",{"metricId":65,"metricName":66,"metricCode":67,"maxScore":27,"score":23},4,"代码能力","coding",{"metricId":15,"metricName":69,"metricCode":70,"maxScore":27,"score":23},"多模态","multimodal",{"metricId":72,"metricName":73,"metricCode":74,"maxScore":27,"score":23},6,"长文本","long-context",{"metricId":76,"metricName":77,"metricCode":78,"maxScore":27,"score":79},7,"性价比","cost-performance",8.5,[81,83,85,89,93,94,95,98],{"id":15,"name":16,"color":17,"type":18,"toolCount":82},35,{"id":27,"name":28,"color":29,"type":18,"toolCount":84},33,{"id":34,"name":86,"color":87,"type":18,"toolCount":88},"免费增值","#409EFF",21,{"id":65,"name":90,"color":91,"type":18,"toolCount":92},"开源","#909399",18,{"id":31,"name":32,"color":33,"type":18,"toolCount":27},{"id":23,"name":24,"color":25,"type":18,"toolCount":27},{"id":61,"name":96,"color":97,"type":18,"toolCount":76},"付费使用","#E6A23C",{"id":10,"name":20,"color":21,"type":18,"toolCount":76}]