这些是读者照片,与图钉带来源的媒体分开显示。
第一项始终是图钉的来源。总体可信度是各份资料对上方所用开始和结束时间支持程度的加权平均;资料每比最新的一份旧 180 天,权重减半。
阿里巴巴的新闻稿署杭州、日期“May 20”,称“Unveiled at the Alibaba Cloud[5] Summit, Qwen3.7-Max is Alibaba's latest large language model”;TechNode[3]报道称其于“on Wednesday”(5月20日)发布,Model Studio将发布时快照列为qwen3.7-max-2026-05-20。Qwen自家博客页面标注日期为2026/05/19,阿里云社区转载于5月21日。
Model Studio page: qwen3.7-max is the largest model in the Qwen3.7 series, text-only at launch; snapshots qwen3.7-max-preview (thinking only), qwen3.7-max-2026-05-20 and qwen3.7-max-2026-06-08 (adds image and video input); 1,000,000-token context, 131,072 max output, 262,144 chain-of-thought; CNY 12 input and CNY 36 output per million tokens in Beijing.[1]
TechNode: Alibaba's Qwen[1] team unveiled Qwen3.7-Max "on Wednesday", its most advanced agent model to date, with API access "set to roll out soon"; a kernel-optimisation test used 1,000+ tool calls and ran up to 35 hours.
OpenRouter lists Qwen[1]: Qwen3.7 Max as text in and out, 1.0M context, $1.475 per million input and $4.425 per million output tokens, released 21 May 2026, with explicit prompt caching.
Alibaba's summit release: Qwen3.7-Max "unveiled at the Alibaba Cloud Summit" on 20 May, engineered for agentic coding, complex reasoning and long-horizon tasks, sustaining up to 35 hours and 1,000+ tool calls, "available soon" on Model Studio; launched alongside T-Head's Zhenwu M890 chip and the Panjiu AL128 supernode server.
有遗漏或错误吗?用你自己的话说明:能佐证此图钉的链接、不同的开始或结束日期及理由,或缺失、有误的信息。AI 会对照此图钉的来源进行核实,搜索更好的来源,并添加任何支持你说法的页面。图钉自身的来源仍然最重要。AI 也会查看图片:显示的是别的东西或显示效果差的图片会被移到后面或替换。
原标题: Qwen3.7-Max Released