█

AI/TLDR

Buy me a coffee
Buy me a coffee

Qwen4-Max

Alibaba's first Qwen4 flagship — a multimodal MoE with native vision and a 1M-token context, at $1.50 / $4.50 per million tokens.

IN SHORT

What is Qwen4-Max?

Qwen4-Max is the general release of Alibaba's Qwen-Max flagship for the Qwen4 generation, announced on 16 November 2026 after a Token Plan-only preview that opened on 11 November. Qwen Cloud's documentation calls it the first model in the Qwen4 family, combining reasoning, native vision and long-running task execution. It has a 1M-token context window and costs $1.50 per million input tokens and $4.50 per million output tokens.

Older version
Qwen4-Max-Preview · 11 November 2026
Where to use it
QwenCloud
Updated Checked against 3 sourcesMarkdown version ↗
Cite this pageAI/TLDR, "Qwen4-Max: specs, pricing and benchmarks", https://ai-tldr.dev/models/qwen4-max/ (updated 16 November 2026)
// FACTSas of 16 November 2026
Context window1M[1]
Input price$1.50 / 1M tokenscached $0.075[2]
Output price$4.50 / 1M tokens[2]
LicenceProprietary (hosted API) · API only[1]
Released16 November 2026[3]
ParametersUndisclosed[1]
  1. [1] docs.qwencloud.com Qwen Cloud docs — Latest model: Qwen4-Max
  2. [2] qwencloud.com QwenCloud — Qwen4-Max model page and pricing
  3. [3] qwen.ai Qwen blog — Qwen4-Max
Qwen-MaxAPI onlyGenerally available
Released
16 Nov 2026
Context
1M
Parameters
Undisclosed
Input
$1.50 / 1M tokens
License
Proprietary (hosted API)
Coverage
2 stories

Overview

Qwen4-Max is the general release of Alibaba's Qwen-Max flagship for the Qwen4 generation, announced on 16 November 2026 after a Token Plan-only preview that opened on 11 November. Qwen Cloud's documentation calls it the first model in the Qwen4 family, combining reasoning, native vision and long-running task execution. The Qwen team presents it as an AI teammate for work, creative tasks and scientific discovery, with examples from 3D modeling and game development to manufacturing, finance, law and healthcare.

The model is built on a new Mixture-of-Experts architecture. It accepts text, images and video, returns text, and has a 1M-token context window. Thinking is on by default with reasoning_effort set to high, and the effort can be set to low, high or xhigh. The endpoint is compatible with OpenAI Chat Completions, OpenAI Responses and Anthropic Messages, and supports function calling, structured output, context caching, batches and web search, plus built-in tools such as a code interpreter and web extractor.

Alibaba says Qwen4-Max outperforms Qwen3.8-Max while using up to about 50% fewer tokens, and that average cost per task drops by more than 60% in its internal evaluation; no independent benchmark had confirmed this at launch. It is also cheaper per token than Qwen3.8-Max: $1.50 input and $4.50 output per million tokens, against $2 and $6. The Qwen team says open weights of smaller Qwen4 models will follow, without a date.

Released2026-11-16
LicenseProprietary (hosted API)
WeightsAPI only
ParametersUndisclosed
Context1M
ArchitectureMixture-of-Experts — Alibaba describes a new MoE architecture with native vision.
ModalitiesText, Image, Video
StatusGenerally available

Pricing

Input$1.50 / 1M tokens
Cached input$0.075 / 1M tokens
Output$4.50 / 1M tokens

Implicit cache input and explicit cache reads are $0.075 per 1M tokens; explicit cache creation is $1.875 (5-minute) or $3 (1-hour) per 1M tokens.

Pricing source ↗

Strengths

  • 1M-token context with native image and video input
  • Lower price than Qwen3.8-Max — $1.50 / $4.50 per million tokens, with implicit cache input at $0.075 (70% less)
  • Alibaba reports up to about 50% fewer tokens per task than Qwen3.8-Max
  • One endpoint that speaks OpenAI Chat Completions, OpenAI Responses and Anthropic Messages, so coding agents can use it by changing the model id
  • Adjustable reasoning effort (low, high, xhigh) with thinking on by default

Best for

  • Reach for it for long-running coding and office agents where cost per task matters
  • Reach for it for multimodal work over screenshots, documents and video
  • Reach for it to swap a cheaper frontier model into tools built on the OpenAI or Anthropic API formats
  • Note: quality claims are Alibaba's own internal evaluation at launch; check independent results before relying on them

How to access

ProviderModel ID
QwenCloud ↗qwen4-max

Qwen-Max — every version

The full lineage of the Qwen-Max line, newest first. Every version has its own page — click any to compare specs, benchmarks and pricing.

VersionReleasedContextLicense
Qwen4-Maxcurrent2026-11-161MProprietary
Qwen4-Max-Preview2026-11-111MProprietary
Qwen3.8-Max-09022026-09-021MProprietary
Qwen3.8-Max2026-08-031MProprietary (weights published as Qwen3.8-2.4T-A95B)
Qwen3.8-Max-Preview2026-07-19—Proprietary
Qwen3.7-Max2026-05—Proprietary
Qwen3-Max2025-09—Proprietary
Qwen2.5-Max2025-01-29—Proprietary

FAQ

How much does Qwen4-Max cost?

Qwen4-Max costs $1.50 per million input tokens and $4.50 per million output tokens on QwenCloud, with implicit cache input at $0.075 per million tokens. It is also available through Token Plan Personal and Team.

Is Qwen4-Max open-weight?

No. Qwen4-Max is served only through Alibaba's API and apps. The Qwen team says open weights of smaller Qwen4 models will be released later, without a date.

How does Qwen4-Max compare with Qwen3.8-Max?

Alibaba says Qwen4-Max outperforms Qwen3.8-Max with up to about 50% fewer tokens and over 60% lower average cost per task in its internal evaluation. Its token prices are also lower: $1.50 / $4.50 against $2 / $6.