🔍 Read the full analysis: The Qwen4 Max Review: A Better Score, Lower Spend, And Tough Competition on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Alibaba launched Qwen4 Max on Nov. 16, 2026, describing it as a 2.4-trillion-parameter model with vision capabilities and a one-million-token context window. Artificial Analysis gave it a score of 43, up from 34 for its predecessor, and measured a lower cost per index task: $3.27 versus $10.02. It ranked ninth among 10 models on its comparison page; speed data and access to downloadable weights were unavailable at launch.
Alibaba launched Qwen4 Max on Nov. 16, presenting it as a new flagship AI model with a 2.4-trillion-parameter backbone, image and video input, and a one-million-token context window. Independent benchmarking firm Artificial Analysis scored it 43 on its Intelligence Index, up from 34 for the previous Qwen flagship, while estimating a lower cost per index task; the model still ranked ninth among 10 models listed on its comparison page.
Artificial Analysis evaluated Qwen4 Max at its highest reasoning setting, called xhigh. The company placed the score above the 41-point median for models in a similar price range. On the 10-model Qwen4 Max comparison page, however, the model was ninth. The top entry scored 65, and the strongest Chinese model shown, Kimi K3.1 Pro, scored 46. These are index results, not a universal measure of performance across every use.
Artificial Analysis estimated that a run of its index tasks cost $3.27 for 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 independent cost estimate refers to the benchmark tasks, while per-token prices are the provider’s listed rates; they are different measures.
Alibaba also reported improvements on its own Work360 professional-task benchmark: a score of 52.69, up from 47.74 for Qwen3.8 Max, and average cost per task falling from $4.43 to $1.67. The company said its secure-coding results improved and described manufacturing and other use cases. Those figures and examples come from Alibaba, not an independent evaluation. The company has not released model weights, and Artificial Analysis had no speed or latency results on launch day.
Qwen4 Max · launch day
From launch claim to independent check
Alibaba’s numbers come first. An outside measurement shows where the model actually lands.
- The launchAlibaba calls Qwen4 Max a 24/7 “teammate”, with 2.4 trillion parameters and a one-million-token context window.
- The vendor’s own testOn its in-house Work360 benchmark, Alibaba reports 52.69, up from 47.74 for Qwen3.8 Max.
- The independent indexArtificial Analysis scores it 43, up from 34 and just above the median of 41 for models at a similar price.
- 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
Lower Benchmark Costs, Mixed Ranking
The launch presents a trade-off for organizations choosing AI models: Qwen4 Max improved on its predecessor and had a lower measured cost per index task, but its score remained behind most of the models on the cited comparison page. Buyers weighing price against quality may find the cost result relevant, while the ranking shows that a generational improvement does not by itself establish leadership over competing systems.
Practical value will depend on how the model performs on a customer’s own tasks, what reasoning setting is used, and the full cost of API use. Artificial Analysis tested at xhigh reasoning effort, whereas Alibaba’s API default is high, according to the source material. Results from the highest setting may not represent ordinary default use, and the available figures do not establish the model’s speed or its cost on a particular workload.
Access is another factor. Qwen4 Max is offered through Alibaba’s cloud API and apps, but its weights are not available to download. Teams that need to run or adapt a model themselves cannot judge it as they might an open-weight release. Alibaba said smaller Qwen4 models would be open-sourced, but gave no release date.
Popular right now

KitchenAid Artisan 5-Quart Tilt-Head Stand Mixer, Almond Cream

Vitamix 5200 Blender with 64 oz Container, Black

Vitamix Touch & Go Advance 32-Ounce Bar Blender, White
As an affiliate, we earn on qualifying purchases.
From Qwen3.8 to Qwen4 Max
Qwen4 Max is the first model in Alibaba’s new Qwen4 family. The company describes it as a sparse mixture-of-experts system with 2.4 trillion backbone parameters, 71 billion activated per token, and an additional 328 billion N-gram Embedding parameters. Alibaba says the architecture combines Gated DeltaNet layers with Qwen Sparse Attention and other changes. Since the weights have not been published, outside researchers cannot independently inspect those design claims or reproduce the training results.
The comparison with Qwen3.8 Max (0902) is the clearest available measure of change: Artificial Analysis scored the older model 34 and Qwen4 Max 43. The index combines nine evaluations covering areas such as coding, office tasks, knowledge and physics. That makes it a broad benchmark, but it does not substitute for testing a model on a specific product or workflow.
Alibaba’s launch material also cites results from coding tests, including DeepSWE v1.1 and Terminal-Bench 4.0. The source article reports those figures through The Next Web, which in turn cited FourWeekMBA, and notes that the comparison chart in Alibaba’s launch post was an image. Those reported figures should be treated separately from the Artificial Analysis index results: they are not the same test, and some claims originate with Alibaba.
“Your new teammate has arrived.”
— Alibaba’s Qwen team, in its launch announcement
Speed, Weights and Real-World Results
Speed and latency were not available in Artificial Analysis’s launch-day data, so the benchmark results do not show how quickly Qwen4 Max responds or completes work. The model’s weights are also unavailable, preventing outside reproduction of Alibaba’s technical claims. Alibaba says smaller Qwen4 models will be open-sourced, but has not announced when.
Independent evidence is still limited to the reported benchmark snapshot. Alibaba’s Work360, coding and customer-case figures are vendor claims, and the source material does not establish that the reported examples were controlled tests. It is also unclear how Qwen4 Max performs at the API’s default high reasoning setting compared with the xhigh setting used for the Artificial Analysis index, or how the quoted task costs would translate to different user workloads.
More Testing and Open-Model Plans
Further independent evaluations, especially results on latency, task-specific accuracy and default API settings, would help clarify how Qwen4 Max performs outside the index’s highest-reasoning test. Users can also compare its published API rates with their own token volumes and workloads; the index-task estimate alone cannot predict an individual deployment bill.
Alibaba has said it plans to release smaller Qwen4 models as open source, but has not supplied a timetable. Until those models or Qwen4 Max’s weights become available, independent researchers will have limited ability to reproduce the company’s architectural and training claims. The next useful milestones are additional third-party testing and a dated announcement about the smaller models.
Key Questions
What is Qwen4 Max?
Qwen4 Max is Alibaba’s flagship Qwen4 model, launched Nov. 16, 2026. Alibaba describes it as a 2.4-trillion-parameter-backbone model that accepts text, images and video and has a one-million-token context window.
How did it score against its predecessor?
Artificial Analysis gave Qwen4 Max a score of 43 on its Intelligence Index, compared with 34 for Qwen3.8 Max (0902). It tested Qwen4 Max at xhigh reasoning effort.
Is Qwen4 Max cheaper to use?
Artificial Analysis estimated $3.27 per index task, compared with $10.02 for the predecessor. Alibaba’s listed API prices are $1.50 per million input tokens and $4.50 per million output tokens. Actual costs depend on usage, and these figures measure different things.
Can users download Qwen4 Max?
No downloadable weights were available at launch. Alibaba said smaller Qwen4 models would be open-sourced but did not give a release date.
How does Qwen4 Max compare with top models?
It ranked ninth among 10 models on the Artificial Analysis page cited in the source material. Its score was above the 41-point median for models in a similar price range, but below the page’s top score of 65. The result is specific to that index and comparison set.
Source: ThorstenMeyerAI.com
