Alibaba opens a preview of Qwen4-Max

PC Watch reported that Qwen, Alibaba’s model development team, announced Qwen4 and began offering Qwen4-Max as a preview on Nov. 11, 2026, Japan time.[2] TechNode described Qwen4-Max as the flagship model in Alibaba’s next-generation Qwen family, while QwenCloud calls the preview a next-generation flagship multimodal model.[1][3] Qwen says Qwen4 is built for complex coding, knowledge work, front-end builds and professional software workflows.[2] These are the company’s stated target uses; the available evidence does not establish how well the preview performs on those tasks.[2]

QwenCloud says the model accepts text, images and video, including for screenshot, document and video analysis, and combines text generation with image and video understanding and deep reasoning.[3] It also supports function calling, web search and built-in tools, alongside structured outputs in JSON Object and JSON Schema formats.[3] The listed context length is 1 million tokens, and the maximum output is 393,216 tokens shared between reasoning and the final answer.[3] Users can choose low, medium or xhigh reasoning effort, with xhigh set as the default.[3]

QwenCloud lists compatibility with OpenAI Chat Completions, Responses and Anthropic Messages APIs.[3] That gives technical teams specific interfaces to examine when assessing integration, but the compatibility listing alone does not establish equivalent behavior or performance.[3][2] As of Nov. 11, PC Watch reported that Qwen had not disclosed Qwen4-Max’s parameter count, benchmark scores or formal release timing.[2] PC Watch also said Qwen had not shown how the model would exceed the scale of Qwen3.8-Max.[2] Teams should therefore test against their own workloads rather than assume a performance advantage.[2]

QwenCloud says the preview is available through Token Plan only, with no pay-as-you-go access, and runs from 10:00 on Nov. 11 to 24:00 on Nov. 25, 2026, UTC+8; it says the model will be retired afterward.[3] TechNode and PC Watch also reported access through Qwen Studio, QwenWork and Qoder, while TechNode additionally listed the Qwen app.[1][2] PC Watch reported monthly plan prices of ¥690, ¥2,890 and ¥16,800, with model and agent usage limits varying by plan.[2] It also reported an event-period offer for new users starting at $3 per month.[2] Token Plan connects users to Qwen and other models, including DeepSeek and GLM, through one subscription, according to PC Watch.[2]