Skip to main content
Rase AI Intelligence Newsletter

Models & assistants Model or AI service

Alibaba releases Qwen4-Max for long multimodal inputs

Alibaba released Qwen4-Max through its hosted products. It accepts text and visual inputs and has a published context window of one million tokens. [1][3]

Model access
Hosted Qwen4-Max in Alibaba Cloud Model Studio. A separate preview endpoint was offered earlier. [1][2]
License
Provider terms apply to hosted access. [1]

What the model does

A separate preview began on November 11, 2026 under a restricted provider plan. Alibaba announced the general Qwen4-Max model on November 16. [2][3]

Why it matters

The longer input window lets a team ask questions across several related files in one request. Visual inputs also allow it to consider information such as a chart alongside the written text. [1]

Who can use it

Developers can evaluate the hosted model in Alibaba Cloud Model Studio. Access options depend on the model ID and the provider’s account and region rules. [1][3]

Key dates

  1. Restricted Qwen4-Max preview begins [2]
  2. Alibaba announces general Qwen4-Max [1][3]

Ways to use it

For a set of long documents, test whether the model uses the relevant source passage and compare visual answers with the original charts. Record the endpoint version. [1]

Model limitations

The million-token limit describes context capacity. Output limits and access conditions vary by endpoint, and the preview is separately identified from the general release. [1][2]

Sources and verification

Catalog evidence checked How we verify records

  1. Primary source Alibaba Cloud model documentation for the general Qwen4-Max release
  2. Primary source Alibaba Cloud’s Qwen4-Max restricted preview dates and terms
  3. Primary source Alibaba Unveils Qwen4-Max: New AI Teammate for Work, Creativity, and Discovery - Alibaba Cloud
  4. Independent source Qwen4 Max - Intelligence, Performance & Price Analysis | Artificial Analysis

Related profiles

Models & assistants

Hy4

Tencent (Hunyuan)

It is a language model family with documented preview and subsequent variants.

Medium confidenceEvidence checked

Models & assistants

Gemini 4 Hydrogen

Google DeepMind

It processes text and other supported inputs for tasks such as code assistance and planning.

High confidenceEvidence checked

Models & assistants

ERNIE 5.5 Flash Preview

Baidu

It is designed to process supported media inputs alongside text in the documented preview environment.

High confidenceEvidence checked