Nine Points Better For Less: The Qwen4 Max Score And Cost Breakdown
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: Nine Points Better For Less: The Qwen4 Max Score And Cost Breakdown on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get smart everyday buys delivered free with Prime

  • Fast, free delivery on millions of items
  • Prime Video, Amazon Music and more included
  • Member-only deals all year
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

TL;DR

Artificial Analysis gave Alibaba’s Qwen4 Max an Intelligence Index score of 43, nine points above its predecessor, and estimated a $3.27 cost per index task versus $10.02 for the older model. The new model ranks ninth among 10 on its Artificial Analysis page; speed data are unavailable, and its weights cannot be downloaded.

Alibaba’s Qwen4 Max scored 43 on the Artificial Analysis Intelligence Index, nine points above its predecessor, while its estimated cost for an index task was $3.27, compared with $10.02 for the earlier model. The independent benchmark’s first results, published November 16, also place Qwen4 Max ninth among 10 models listed on its page, tempering Alibaba’s launch claims about the new flagship.

Artificial Analysis tested Qwen4 Max using its Intelligence Index v5.0.1, which combines nine evaluations spanning areas such as agentic coding, office work, knowledge and physics. The model was tested at its highest reasoning setting, xhigh. Alibaba’s API default is high, so results from a default configuration may differ. Artificial Analysis describes the score as above the median of 41 among comparable models.

The benchmark’s model-page comparison puts Qwen4 Max at 43 and Qwen3.8 Max (0902) at 34. The top score shown is 65. The estimated cost per index task is $3.27 for the new model and $10.02 for its predecessor. Alibaba’s listed API prices are $1.50 per million input tokens and $4.50 per million output tokens; actual charges depend on usage.

Alibaba separately reports improvements on its own Work360 test: Qwen4 Max scored 52.69, up from 47.74 for Qwen3.8 Max, and the company says average task cost fell from $4.43 to $1.67. Those are vendor-reported figures, not independent benchmark results. Alibaba also reports that its secure-coding measure rose from 22.0% to 31.8% of functions passing both functionality and security checks.

At a glance
reportWhen: Published November 16, 2026; initial be…
The developmentArtificial Analysis published its first Qwen4 Max results on November 16, showing a higher score and lower estimated task cost than the previous Qwen flagship.

Qwen4 Max · launch day

From launch claim to independent check

Alibaba’s numbers come first. An outside measurement shows where the model actually lands.

Alibaba: 2.4T parameters, 71B active Own test Work360: 47.74 → 52.69 Independent index (0 to 70) 0 70 34 41 43 65 Speed: no data yet
  1. The launchAlibaba calls Qwen4 Max a 24/7 “teammate”, with 2.4 trillion parameters and a one-million-token context window.
  2. The vendor’s own testOn its in-house Work360 benchmark, Alibaba reports 52.69, up from 47.74 for Qwen3.8 Max.
  3. The independent indexArtificial Analysis scores it 43, up from 34 and just above the median of 41 for models at a similar price.
  4. What is still openThe top model on the same page scores 65. Speed is not measured yet, and the weights are closed.

Sources: Alibaba/Qwen launch blog, Nov 16, 2026 (vendor figures); Artificial Analysis, model pages for Qwen4 Max and Qwen3.8 Max (0902), retrieved Nov 16, 2026. Schematic scale; marker positions follow the index values.

Graphic: Powered by Thorsten Meyer AI

A Better Score at Lower Cost

The early findings suggest Alibaba has improved its flagship model while lowering the measured cost of running the benchmark workload. That combination matters to developers and businesses weighing model quality against API expense. The $3.27 figure is for an Artificial Analysis index task, however; it is not a universal price for every user request, and workloads can vary in token use and reasoning demands.

The ranking also puts the improvement in perspective. A score of 43 is above the comparable-model median, but Qwen4 Max sits ninth of 10 on its page, below the leading models listed there. The results support a claim of progress over the previous Qwen flagship, not a claim that Qwen4 Max leads the wider field.

Alibaba’s lower costs may make the model worth testing for high-volume applications, but buyers still need evidence about latency, reliability and performance on their own tasks. The benchmark does not answer those operational questions, and the model’s closed weights limit independent inspection and on-premises deployment.

Products Worth Considering

Logitech MX Keys S Combo Wireless Keyboard and Mouse, Graphite

Logitech MX Keys S Combo Wireless Keyboard and Mouse, Graphite

Logitech MX Ergo S Advanced Wireless Trackball Mouse USB-C - Graphite

Logitech MX Ergo S Advanced Wireless Trackball Mouse USB-C – Graphite

As an affiliate, we earn on qualifying purchases.

Alibaba’s New Qwen Flagship

Alibaba introduced Qwen4 Max on November 16 as the first model in its Qwen4 family, calling it a “24/7 AI teammate” for work, creativity and discovery. The company says the model has a 2.4-trillion-parameter backbone, 71 billion parameters activated per token and an additional 328 billion N-gram embedding parameters. It describes the system as a sparse mixture-of-experts model, in which only part of the network is active for each token.

According to Alibaba’s documentation, Qwen4 Max handles text, images and video and generates text. It is offered through the QwenCloud API, a Token Plan subscription and Alibaba applications including QwenWork, Qoder and QwenStudio. The source reports that a limited preview began November 11.

Alibaba also describes changes to its attention and residual-branch design and says training completed with no loss spikes. Those architecture and training statements come from the company; because the model weights are not public, outside researchers cannot independently examine them. Alibaba says smaller Qwen4 models will be open-sourced but has not provided a release date.

““In our Work360 evaluation, the average cost per task drops from $4.43 to $1.67 — a 62% reduction.””

— Alibaba

Speed, Weights and Real-World Results

Artificial Analysis had no speed or latency data for Qwen4 Max when it published its first results. It is therefore not yet possible to compare how quickly the model responds or completes tasks against its predecessor or competitors using the supplied benchmark information.

The independent index offers an initial standardized comparison, but it does not establish how Qwen4 Max will perform across every business workload. Alibaba’s Work360 results and demonstrations—including claims about manufacturing, legal review and other applications—are company-reported cases, not controlled independent tests. The source material does not provide enough detail to verify those demonstrations or assess whether their results generalize.

The weights are not available for download, and Alibaba has not announced a date for releasing smaller open-source Qwen4 models. The available information also does not resolve how performance changes at the API’s high default reasoning setting compared with the xhigh setting used for the index test.

More Tests and Open-Model Plans

The next useful evidence would include Artificial Analysis results for latency and speed, plus further independent testing across coding, business and multimodal tasks. Comparisons using the API’s default high setting would also help users understand what to expect without changing configuration. No publication schedule for those additional measurements is given in the source material.

Alibaba has said that smaller models in the Qwen4 family will be open-sourced, but has not announced when or under what release terms. Until those details and additional test results are available, the confirmed picture is limited to the initial benchmark: a nine-point gain over Qwen3.8 Max, a lower estimated index-task cost and a ranking that remains near the bottom of the 10-model page.

Key Questions

What score did Qwen4 Max receive?

Qwen4 Max scored 43 on Artificial Analysis’s Intelligence Index v5.0.1. It was tested at the xhigh reasoning setting.

How does it compare with the previous Qwen flagship?

Qwen3.8 Max (0902) scored 34 on the same index page, making the reported difference nine points. Artificial Analysis estimated index-task costs of $10.02 for the older model and $3.27 for Qwen4 Max.

Is Qwen4 Max the top-ranked model?

No. It ranks ninth among 10 models listed on its Artificial Analysis page. The highest score shown there is 65.

Can users download Qwen4 Max’s weights?

No. The source says Qwen4 Max is available through Alibaba’s services and cannot be downloaded. Alibaba says smaller Qwen4 models will be open-sourced, but has not set a release date.

Are the reported cost savings independently verified?

Artificial Analysis independently reports an estimated $3.27 cost per index task, compared with $10.02 for Qwen3.8 Max. Alibaba’s separate Work360 cost figures are company-reported and should not be treated as the same measurement.

Source: ThorstenMeyerAI.com

This content is for general information only and is not financial, tax or legal advice. Consult a qualified professional for decisions about your money.
You May Also Like

The Delegation Ladder: The Four Agentic Loops, and What Each One Lets You Stop Doing

An analysis of the four agentic loops in AI engineering, explaining what each allows you to stop doing and how they shape AI processes.

A Checklist For Building An AI Automation Desk In 2026

A 2026 desk setup checklist covers a workstation laptop, optional AI hardware, desk connectivity and portable storage, with compatibility checks before buying.

What To Do If Your Email Was Exposed In The Osaka Data Breach

Osaka says 1,078,713 email addresses were exposed through email service Arara. Here is who may be affected and how to handle suspicious messages.

Could Bots And Math Shift The Horizon For Human Work?

A commentary links persistent AI agents, automated workflows and a reported Navier–Stokes result to questions about how work may be organized.