🔍 Read the full analysis: Qwen4 Max Takes A Step Forward While Staying Behind Top Competitors on ThorstenMeyerAI.com
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
TL;DR
Alibaba launched Qwen4-Max on November 16, 2026, describing a 2.4-trillion-parameter model with vision and a one-million-token context window. Artificial Analysis gave it an Intelligence Index score of 43, up from 34 for its predecessor, while ranking it ninth among 10 models on the Qwen4-Max comparison page. Its measured index-task cost was lower than the previous model’s; speed data and downloadable weights were not available at launch.
Alibaba launched Qwen4-Max on November 16, 2026, as the first model in its new Qwen4 family. Same-day results from independent benchmarking company Artificial Analysis put the model at 43 on its Intelligence Index, nine points above the previous Qwen flagship, but ninth among 10 models listed on Qwen4-Max’s comparison page.
Artificial Analysis says Qwen4-Max scored above the 41-point median for models in a similar price range. Its page lists a top score of 65; other high-ranking models named in the source include Claude Fable 5.2 at 65, Gemini 4 Carbon at 60, Claude Opus 5.5 at 59, GPT-6 Astra at 57 and GPT-6.1 Sol at 55. The strongest Chinese model identified on that chart, Kimi K3.1 Pro, scored 46. These figures describe the specific benchmark comparison, not performance on every task.
The measured cost for one Artificial Analysis index task was $3.27 for Qwen4-Max, compared with $10.02 for Qwen3.8 Max (0902). Alibaba’s published API list prices are $1.50 per million input tokens and $4.50 per million output tokens. Artificial Analysis tested the model at its highest reasoning setting, xhigh; the source says Alibaba’s API default is high, so results under default settings may differ.
Alibaba says Qwen4-Max has 2.4 trillion backbone parameters, with 71 billion activated per token and an additional 328 billion N-gram Embedding parameters. The company describes it as able to process text, images and video and generate text, with a one-million-token context window. It is available through QwenCloud’s API, a subscription called Token Plan, and Alibaba products QwenWork, Qoder and QwenStudio. The weights are not available to download.
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
The Benchmark Gain Comes With Trade-Offs
The results show a measurable improvement over Alibaba’s preceding flagship without placing Qwen4-Max among the highest-scoring models in the comparison. For users and organizations choosing an AI model, that distinction matters: a nine-point rise is evidence of progress on this index, while the ninth-place ranking cautions against treating the launch as a broad lead over competitors.
Price may be part of the appeal. Artificial Analysis’s reported $3.27 per index task is below the previous model’s $10.02 on the same measure. That could matter to developers or businesses running repeated tasks, but a single benchmark’s task cost does not establish total operating cost for every workload. Actual bills can depend on input and output length, reasoning settings, API terms and the tasks being run.
Access also shapes the decision. Qwen4-Max is offered through Alibaba services, but its closed weights prevent users from downloading and independently running the model. That limits outside inspection and local deployment compared with models whose weights are published. Alibaba says smaller Qwen4 models will be open-sourced, but has not given a date in the supplied material.
Popular right now

Bose QuietComfort Headphones, White Smoke

Ninja SP151 Flip Toaster Oven & Air Fryer, Stainless Steel

De'Longhi Classic Signature Espresso Machine with Adjustable Milk Frother
As an affiliate, we earn on qualifying purchases.
How Qwen4-Max Was Evaluated
Artificial Analysis’s Intelligence Index v5.0.1 combines nine evaluations covering areas such as agentic coding, office automation, knowledge and physics. The company says Qwen4-Max was tested at xhigh reasoning effort. Its result is therefore tied to that benchmark version and setting; it should not be read as a direct prediction of performance under every API configuration.
Alibaba also published its own Work360 results. The company says Qwen4-Max scored 52.69, up from 47.74 for Qwen3.8-Max, on a professional-task benchmark it describes as spanning 26 top-level domains, 182 subdomains and 721 specialized fields. Alibaba reports that average task cost on Work360 fell from $4.43 to $1.67. These are vendor-reported figures, rather than results from the independent Artificial Analysis evaluation.
The company’s launch materials also cite improvements on coding and secure-coding tests and describe industrial and professional-use demonstrations. Those claims have different evidentiary status from the independently reported index result: the source characterizes the examples as vendor case studies, not controlled tests. Reports citing Alibaba’s comparison chart give mixed results across coding benchmarks, and the chart itself is presented as an image in the launch materials.
““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
Speed and latency figures were not available from Artificial Analysis on launch day, according to the source. That leaves an important practical question unanswered: whether the model’s lower measured task cost comes with comparable response times to competing systems.
Because Qwen4-Max’s weights are not published, outside researchers cannot inspect or independently run the model’s architecture and training claims. Alibaba describes its attention design and says training proceeded without loss spikes, but those details cannot be checked from weights that are unavailable. The company has said smaller Qwen4 models will be open-sourced, with no release date specified in the supplied material.
It is also unclear how the benchmark ranking translates to specific customer workloads. The independent index is one composite measure, while Alibaba’s Work360 figures and customer examples come from company materials. The source does not provide enough information to independently verify the case-study outcomes or to establish whether they generalize to other users.
More Independent Testing Needed
The next useful evidence will be additional independent results, particularly on speed, latency and task-specific performance, alongside repeatable comparisons using the same reasoning settings and model versions. Buyers should also check current API pricing and availability directly with Alibaba, as service terms can change after launch.
Alibaba has said smaller models in the Qwen4 family will be open-sourced, but no timetable is provided in the source. Until that happens, Qwen4-Max remains an API- and product-access model rather than a downloadable release. Its position could become clearer as benchmark coverage expands and more users report results on their own workloads.
Key Questions
What did Artificial Analysis report for Qwen4-Max?
It reported an Intelligence Index score of 43, compared with 34 for Qwen3.8 Max (0902). Qwen4-Max ranked ninth among the 10 models on its comparison page and was above the listed median of 41 for models in a similar price range.
Is Qwen4-Max ahead of the leading competitors in the reported comparison?
No. The source lists several models with higher scores, including Claude Fable 5.2 at 65, Gemini 4 Carbon at 60 and Claude Opus 5.5 at 59. The ranking applies to the specific Artificial Analysis page and index, not every possible use case.
How much does Qwen4-Max cost to run?
Artificial Analysis reported $3.27 per index task, compared with $10.02 for the predecessor on that evaluation. 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 task length.
Can users download Qwen4-Max?
No. The source says Qwen4-Max has closed weights and is available through Alibaba’s API and products. Alibaba says smaller Qwen4 models will be open-sourced but has not announced a release date.
What important information is still missing?
Artificial Analysis had no speed or latency data for the model on launch day. Independent verification of Alibaba’s demonstrations and some vendor-reported benchmark claims is also limited in the source material.
Source: ThorstenMeyerAI.com
