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

TL;DR

Alibaba launched Qwen4 Max on Nov. 16, 2026, and Artificial Analysis published an independent score of 43, nine points above its predecessor. The service costs less per benchmark task than the prior model, but ranks ninth among 10 models on its Qwen4 Max page; speed data and downloadable weights are not available.

Alibaba launched Qwen4 Max on Nov. 16, 2026, and independent benchmarking firm Artificial Analysis scored it 43 on its Intelligence Index, up from 34 for the previous Qwen flagship. The result places the model above the index median of 41 for similarly priced models, but it ranks ninth among 10 models listed on the Qwen4 Max page, while costing less per benchmark task than its predecessor.

Buying for a business?Offer from Amazon

Get business pricing on tech for your team

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

Artificial Analysis reports that a task in its index costs $3.27 with Qwen4 Max, compared with $10.02 for Qwen3.8 Max (0902). Alibaba lists API prices of $1.50 per million input tokens and $4.50 per million output tokens. The company also reports lower costs on its own Work360 evaluation, but those figures come from a vendor-run test rather than an independent benchmark.

On Work360, Alibaba says Qwen4 Max scored 52.69, compared with 47.74 for Qwen3.8 Max. The company says average cost per task on that test fell from $4.43 to $1.67. Those results are not directly interchangeable with Artificial Analysis’s index score or task-cost estimate: the tests and their measurement methods differ.

Alibaba describes Qwen4 Max as a 2.4-trillion-parameter model, with 71 billion parameters activated per token, and says it supports text, images and video, as well as a one-million-token context window. The model is available through Alibaba’s QwenCloud API and selected company products. Its weights are closed, so users cannot download and inspect or run the model independently.

At a glance
reportWhen: Launched Nov. 16, 2026; independent res…
The developmentAlibaba launched Qwen4 Max, with launch-day results showing an independent score increase over its predecessor and lower benchmark task costs, alongside a low ranking against leading models.
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.

Lower Costs, Mixed Standing

The launch gives businesses a new option to compare on measured task cost as well as model quality. Artificial Analysis’s estimate suggests Qwen4 Max costs less per index task than its predecessor, while its score rose from 34 to 43. That combination may matter to organizations weighing model performance against recurring API expenses.

The ranking also sets limits on the launch-day result. A score of 43 is above the listed median of 41, but the model sits ninth of 10 on its page; the top score there is 65. The index is one comparison tool, and the source material does not establish how Qwen4 Max performs across every workload. The lower benchmark cost alone does not show that it will be cheaper or better for a particular company’s tasks.

Products Worth Considering

LG UltraGear 27-inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDR True Black 400

LG UltraGear 27-inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDR True Black 400

As an affiliate, we earn on qualifying purchases.

A Closed First Qwen4 Model

Qwen4 Max is the first model in Alibaba’s new Qwen4 family. Alibaba says the architecture combines Gated DeltaNet layers with a method it calls Qwen Sparse Attention for long inputs, alongside other design changes. The company also says training completed without loss spikes. Because the weights are unpublished, outside researchers cannot independently examine the architecture or verify those training claims.

The launch post presents Work360 scores, token-use estimates and examples from coding, manufacturing and other fields. These are Alibaba-reported results and case studies; the provided material does not describe controlled independent tests for them. A limited preview reportedly opened Nov. 11, according to The Next Web. Alibaba says smaller Qwen4 models will be open-sourced, but has not given a release date.

“Your new teammate has arrived.”

— Alibaba’s Qwen team

Speed and Weights Remain Unknown

Artificial Analysis had no speed or latency data for Qwen4 Max on launch day, leaving response time and throughput unmeasured in the cited independent results. The model’s weights are also closed, limiting outside verification of Alibaba’s architecture and training statements.

The provided source material does not establish how the model performs across a broad set of independently run tasks. Alibaba’s Work360 and showcase results remain company-reported, and the launch-day index comparison covers only the models on the cited page. The timing for open-sourcing smaller Qwen4 models is also unknown.

Awaiting Broader Independent Results

Further independent evaluations may clarify Qwen4 Max’s speed, reliability and results across specific tasks. Buyers can compare its listed API rates and available index score with their own workloads, but the launch-day data does not settle how it will perform in production.

Alibaba has said that smaller Qwen4 models will be open-sourced, without announcing a date. A release would allow outside users to inspect and run those models, though it would not by itself make Qwen4 Max downloadable.

Key Questions

What score did Qwen4 Max receive?

Artificial Analysis scored it 43 on its Intelligence Index. Qwen3.8 Max (0902) scored 34 on the same index.

How does it rank against other models?

It ranks ninth among 10 models on the Artificial Analysis page for Qwen4 Max. The top model on that page scores 65.

Is Qwen4 Max cheaper than its predecessor?

Artificial Analysis estimates that one index task costs $3.27 with Qwen4 Max, versus $10.02 with Qwen3.8 Max (0902). Alibaba separately reports a lower average task cost on its own Work360 test.

Can users download Qwen4 Max?

No. The model’s weights are closed, and Alibaba has not announced a date for open-sourcing smaller models in the Qwen4 family.

Are speed results available?

Artificial Analysis had no speed or latency data for Qwen4 Max on launch day, Nov. 16, 2026.

Source: Thorsten Meyer AI

You May Also Like

The Turbulent AI Era Is Here

Recent AI advancements and controversies mark the start of a turbulent new era, raising questions about regulation, safety, and future impact.

A Week Of Using Codex More Than Claude

A recent week-long usage survey shows increased preference for Codex over Claude among users, highlighting shifts in AI tool adoption.

Glasspane: One Dataset, Three Views

Glasspane introduces a demo showcasing a single dataset presented through role-specific views, emphasizing transparency and trust in infrastructure monitoring.

Did Companies Move Too Fast On AI? The Rehire Data Offers Clues

Surveys suggest some employers have rehired workers after AI-related cuts, but the figures are self-reported and do not prove AI caused each reversal.