TL;DR
Alibaba launched Qwen4 Max on November 16, 2026, describing a 2.4-trillion-parameter model with image and video input and a one-million-token context window. Artificial Analysis gave it a score of 43, up from 34 for the previous flagship, and reported a lower cost per index task; it ranked ninth among 10 models on the page. Speed data was unavailable, and Alibaba has not published the model weights.
Alibaba launched Qwen4 Max on November 16, and independent benchmark firm Artificial Analysis scored it 43 on its Intelligence Index, up from 34 for the previous Qwen flagship. The new model also cost $3.27 per index task, compared with $10.02 for its predecessor, but ranked ninth among the 10 models listed on its Artificial Analysis page.
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
Alibaba describes Qwen4 Max as a model with a 2.4-trillion-parameter backbone, 71 billion parameters active per token, and an additional 328 billion N-gram embedding parameters. The company says it accepts text, images and video and has a one-million-token context window. These specifications and the launch description come from Alibaba; the model weights are not available for outside inspection.
Artificial Analysis’s launch-day figures put Qwen4 Max at 43 on a 0-to-70 index. Its predecessor, Qwen3.8 Max (0902), scored 34, while the median for models at a similar price was 41. On the 10-model page, Qwen4 Max was ninth; the leading model scored 65. The benchmark result indicates an improvement over the prior Qwen flagship, while the page ranking places it below most of the listed models.
Artificial Analysis estimated the cost of one index task at $3.27 for Qwen4 Max, against $10.02 for Qwen3.8 Max (0902). Alibaba separately lists API prices of $1.50 per million input tokens and $4.50 per million output tokens. Those per-token prices and the benchmark’s per-task estimate describe different measures, so they should not be treated as interchangeable.

Lower Costs, Ninth-Place Index Rank
The results give organizations weighing model costs and performance a more specific comparison than Alibaba’s launch claims alone. On Artificial Analysis’s measure, Qwen4 Max improved over its predecessor while reducing the reported cost of an index task. That combination may matter to teams running many model tasks, where costs can shape which systems are practical to use.
The ranking also sets a limit on the launch-day case for the model. Its score of 43 was above the comparable-model median of 41, but it remained ninth out of 10 models on the page, well below the leading score of 65. The index is one benchmark, not a complete measure of quality across coding, research, office work or other uses. Buyers still need task-specific results and reliable speed data to judge whether the price difference suits their workloads.
Products Worth Considering

Bose Smart Ultra Dolby Atmos Soundbar

Anker SOLIX C300 Portable Power Station with 60W Solar Panel, 288Wh, 300W

Logitech MX Keys S Combo Wireless Keyboard and Mouse, Graphite
As an affiliate, we earn on qualifying purchases.
Alibaba’s Claims and the Available Checks
Alibaba’s launch announcement presents Qwen4 Max as the first model in its Qwen4 family and as a work-focused, always-available AI assistant. It reports a score of 52.69 on its in-house Work360 test, compared with 47.74 for Qwen3.8 Max, and says average cost per task on that test fell from $4.43 to $1.67. Work360 is Alibaba’s own evaluation, so those results are vendor-reported rather than an independent comparison.
The company also reports reduced token use on Work360 and Terminal-Bench 4.0, improved results on a secure-coding test, and examples involving manufacturing and other professional tasks. These figures and examples are claims in Alibaba’s launch material. Case studies and vendor benchmarks can show what a company says its model achieved, but they do not by themselves establish how it will perform for other customers or under independent testing.
Artificial Analysis published its Qwen4 Max and Qwen3.8 Max (0902) model pages on November 16, 2026. Its launch-day page provided index and cost figures but no speed or latency results. The available benchmark therefore offers an early comparison of measured score and estimated task cost, while leaving important operational questions open.
“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 on launch day, leaving response times and throughput unmeasured in the cited independent results. The model’s weights are closed, so outside researchers cannot inspect or run the full system locally. Alibaba says smaller Qwen4 models will be open-sourced, but the launch material gives no release date.
The available evidence also does not settle how Qwen4 Max performs across different customer tasks, workloads or operating conditions. Alibaba’s Work360 and case-study results are company-reported; Artificial Analysis supplies an independent index score, but one index cannot answer every use-case question. The launch-day figures do not establish how the model compares on speed, reliability or total cost for a particular deployment.
More Independent Tests Needed
The next useful evidence would be speed and latency measurements, alongside further independent results on coding, reasoning and professional tasks. Those comparisons could show whether the lower reported index-task cost comes with tradeoffs that matter in day-to-day use.
Customers can access Qwen4 Max through Alibaba’s QwenCloud API, Token Plan subscription and QwenWork, Qoder and QwenStudio apps, according to the launch material. Alibaba has said smaller Qwen4 models will be open-sourced, but has not set a date. For now, the available record is a launch-day score and cost comparison, with broader performance and release details still pending.
Key Questions
What did Artificial Analysis score Qwen4 Max?
Artificial Analysis gave Qwen4 Max a score of 43 on its 0-to-70 Intelligence Index on November 16, 2026. Qwen3.8 Max (0902) scored 34.
How much does Qwen4 Max cost?
Artificial Analysis estimated $3.27 per index task. Alibaba lists API prices of $1.50 per million input tokens and $4.50 per million output tokens. These are different cost measures.
Is Qwen4 Max a leading model on the index?
It ranked ninth among 10 models on its Artificial Analysis page. The top model there scored 65; Qwen4 Max scored 43.
Can users download Qwen4 Max?
No. The model weights are closed. Alibaba says smaller models in the Qwen4 family will be open-sourced, but has not announced a release date.
Are independent speed results available?
Not in the launch-day Artificial Analysis results cited here. The page had no speed or latency data for Qwen4 Max.
Source: Thorsten Meyer AI
