AIThis post was created with the assistance of artificial intelligence (AI).

TL;DR

Artificial Analysis gave Alibaba’s Qwen4 Max an Intelligence Index score of 43 on Nov. 16, up from 34 for Qwen3.8 Max. The benchmark firm lists the cost at $3.27 per index task, compared with $10.02 for its predecessor, but ranks Qwen4 Max ninth among 10 models on its page. Speed data was unavailable at launch.

Artificial Analysis scored Alibaba’s newly launched Qwen4 Max at 43 on its Intelligence Index on Nov. 16, nine points above its predecessor, while estimating an index task costs $3.27 compared with $10.02 for Qwen3.8 Max. The independent benchmark offers an early measure of the model behind Alibaba’s “24/7 AI teammate” pitch, though its score places it ninth among 10 models listed on the Qwen4 Max page.

Buying for a business?Offer from Amazon

Get business pricing on monitors, keyboards and dev gear

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

The index score for Qwen4 Max is 43 on a 0-to-70 scale, according to Artificial Analysis. Qwen3.8 Max (0902) scored 34 on the same index. The benchmark firm lists the median score for models at a similar price as 41, putting Qwen4 Max slightly above that reference point. The highest-scoring model on the Qwen4 Max comparison page has a score of 65.

Artificial Analysis estimates a cost of $3.27 per index task for Qwen4 Max, versus $10.02 for Qwen3.8 Max. Alibaba’s published API prices are $1.50 per million input tokens and $4.50 per million output tokens. These figures describe different cost measures: the per-task estimate is tied to the benchmark’s workload, while token prices depend on how much a user sends and receives.

Alibaba separately reports that its in-house Work360 score rose from 47.74 to 52.69 and average cost per task fell from $4.43 to $1.67. Those are vendor-reported results, not independent measurements. Artificial Analysis had no speed or latency data for Qwen4 Max on launch day, and its index score does not by itself establish performance across every kind of work.

At a glance
reportWhen: Published Nov. 16, 2026; launch-day mea…
The developmentArtificial Analysis published its first Qwen4 Max results, finding a higher index score and lower cost per task than the previous Qwen flagship.
Bar chart titled Ninth of ten on its own page, nine points above its predecessor. Artificial Analysis Intelligence Index for the 10 models on the Qwen4 Max page: Claude Fable 5.2 (max) 65, Gemini 4 Carbon (high) 60, Claude Opus 5.5 (max) 59, GPT-6 Astra (max) 57, GPT-6.1 Sol (max) 55, Muse Wonder Full (max) 50, Kimi K3.1 Pro (open weights) 46, Muse Wonder (max) 44, Qwen4 Max 43 (highlighted), Grok 4.7 (xhigh) 35. Below a dashed line: the predecessor Qwen3.8 Max (0902) 34
Qwen4 Max is ninth of the ten models on its Artificial Analysis page and nine points above its predecessor. Source: Artificial Analysis, model pages for Qwen4 Max and Qwen3.8 Max (0902), retrieved November 16, 2026. Chart: Thorsten Meyer AI.

A Lower Benchmark Bill

The early results suggest a possible trade-off for businesses comparing model costs with benchmark performance: Qwen4 Max scored above the price-group median cited by Artificial Analysis while costing less per index task than its predecessor. That may make it worth evaluating for workloads where cost matters, but the index and its task-cost estimate are only one comparison point. They do not show what a specific company would pay or how well the model handles its own tasks.

The ranking also tempers the score improvement. Qwen4 Max sits ninth of 10 models on the page, well below the leading score of 65. A higher score than the previous Qwen flagship is a measurable step up on this index; it is not evidence that the model leads the broader field. Missing speed data also leaves a practical procurement question unanswered: how quickly it responds under real usage.

Products Worth Considering

LG UltraGear 32GX870A-B 32-inch 4K UHD OLED Dual-Mode Gaming Monitor

LG UltraGear 32GX870A-B 32-inch 4K UHD OLED Dual-Mode Gaming Monitor

Samsung Galaxy Tab S10 Lite, 10.9-inch LCD, 6GB RAM, 128GB Storage, Gray

Samsung Galaxy Tab S10 Lite, 10.9-inch LCD, 6GB RAM, 128GB Storage, Gray

As an affiliate, we earn on qualifying purchases.

Qwen4 Max at Launch

Alibaba introduced Qwen4 Max as the first model in its Qwen4 family on Nov. 16. The company describes it as a model with a 2.4-trillion-parameter backbone, 71 billion parameters active per token, and a one-million-token context window. It says the model can process text, images and video and produce text. Those specifications and capabilities come from Alibaba’s launch materials.

Alibaba says Qwen4 Max is available through the QwenCloud API, its Token Plan subscription, and the QwenWork, Qoder and QwenStudio apps. The model’s weights are closed, so users cannot download them. Alibaba has said smaller Qwen4 models will be open-sourced but has not announced a release date. A limited preview reportedly opened on Nov. 11, according to The Next Web.

The launch post also reports results from Alibaba’s own Work360 benchmark and describes demonstrations involving coding, manufacturing and other tasks. These are company-reported evaluations and case studies. Artificial Analysis’s index is a separate comparison, but it does not validate every claim in Alibaba’s launch materials.

“Your new teammate has arrived.”

— Alibaba’s Qwen team

Speed and Real-World Costs

Latency and speed remain unmeasured in Artificial Analysis’s launch-day listing. The available cost figures also do not predict each buyer’s bill: API use will vary with input and output volume, and the index-task estimate reflects the benchmark workload.

It is not yet clear how Qwen4 Max performs across a broad set of independently assessed tasks beyond the published index, or how its results change under different settings and workloads. Alibaba’s Work360 figures and case studies have not been independently established by the material available here. The model’s weights are closed, which also limits outside inspection of its architecture and training claims.

More Independent Results Needed

The next useful evidence will be independent testing that reports speed alongside task quality and costs across a wider range of workloads. Buyers can compare the current API prices and benchmark estimates with their own usage, while treating Alibaba’s Work360 results as vendor-reported. The Qwen4 Max page may gain further measurements as testing continues.

Alibaba has said that smaller models in the Qwen4 family will be open-sourced, but has not given a date. Until a release schedule or further benchmark data appears, Qwen4 Max’s launch-day record consists of a higher score than its predecessor, a lower estimated index-task cost, and unresolved questions about speed and performance in specific applications.

Key Questions

What score did Qwen4 Max receive?

Artificial Analysis scored it 43 on its 0-to-70 Intelligence Index on Nov. 16, 2026. Its predecessor, Qwen3.8 Max (0902), scored 34.

How much does a Qwen4 Max benchmark task cost?

Artificial Analysis estimates $3.27 per index task, compared with $10.02 for Qwen3.8 Max. Alibaba lists API prices of $1.50 per million input tokens and $4.50 per million output tokens. Actual spending depends on usage.

Is Qwen4 Max the top-ranked model?

No. On the Artificial Analysis page cited at launch, it ranks ninth among 10 models. The highest score on that page is 65.

Can users download Qwen4 Max?

No. Qwen4 Max’s weights are closed, and Alibaba offers access through its API, subscription and apps. The company says smaller Qwen4 models will be open-sourced, but has not announced when.

Source: Thorsten Meyer AI

You May Also Like

Why SAP’s €1 Billion AI Focus Is On Data Tables, Not Just Chatbots

SAP has acquired Prior Labs, investing over €1 billion to develop advanced tabular AI models, emphasizing structured data over traditional chatbots.

AI Labs’ Suppliers Are Becoming Their Shareholders

TSMC approved up to $500 million each for Anthropic and OpenAI, as seven companies with AI supply-chain ties agreed to buy $26 billion of Anthropic stock.

Operational SOP drift detector for franchise operators

A new SOP comparison tool is being tested for multi-location franchises to detect procedural drift, helping maintain consistency across outlets.

Report: Meta And Microsoft Scale Back Use Of Anthropic’s Claude

A Yahoo Finance headline reports reduced internal use of Anthropic’s Claude at Meta and Microsoft, but offers no details on scope, timing or reasons.