
Alibaba’s latest AI model, Qwen 4 Max, introduces a multimodal approach by integrating text, image, and video understanding alongside advanced reasoning capabilities. With features like a 1 million token context window and a mixture of experts architecture, it aims to support tasks ranging from coding to professional workflows in fields like finance and healthcare. However, its performance has been mixed, with notable strengths in web development but slower response times and higher costs, as reviewed by World of AI.
Explore how Qwen 4 Max handles complex tasks such as generating 3D games in Unity and producing detailed legal and financial reports. Gain insight into its benchmark comparisons, including its modest improvement over Qwen 3.8 Max and its challenges against models like GPT 6.1 Sol. This explainer also previews upcoming additions to the Qwen 4 series, including a smaller 27 billion parameter model expected soon.
Alibaba Launches Qwen 4 Max: A Mixed Bag for AI Enthusiasts
TL;DR Key Takeaways :
- Alibaba has launched Qwen 4 Max, a multimodal AI model with a 1 million token context window, advanced reasoning capabilities, and a mixture of experts architecture, aimed at creative and technical tasks like coding, finance, law, and healthcare.
- Qwen 4 Max reportedly reduces average task costs by over 60% compared to its predecessor, with pricing set at $1.50 per million input tokens and $4.50 per million output tokens, but it remains more expensive per task than GPT 6.1 Sol.
- Performance highlights include strong results in web development and game creation, such as generating a playable 3D game and a Minecraft Dungeons clone, though significant latency issues were noted during testing.
- While Qwen 4 Max shows potential in professional applications like finance, law, and healthcare, its benchmark performance lags behind leading models like GPT 6.1 Sol and Kim K 3.1 Pro, with slower response times and higher costs.
- Alibaba plans to release a smaller 27 billion parameter open-weight model in the Qwen 4 series within two to three weeks, which could address some of Qwen 4 Max’s shortcomings and offer a more competitive option.
Alibaba has introduced its latest AI model, Qwen 4 Max, the first in the Qwen 4 series. This multimodal model integrates text, image, and video understanding with advanced reasoning capabilities. Designed for tasks ranging from coding and web development to professional workflows in finance, law, and healthcare, Qwen 4 Max aims to cater to both creative and technical users. While its specifications are notable, reviews have been mixed, with the model excelling in some areas but underperforming in others.
Key Features and Specifications
Qwen 4 Max offers several advanced features, including a 1 million token context window and a mixture of experts architecture. It reportedly surpasses its predecessor, Qwen 3.8 Max, while using up to 50% fewer tokens. Additionally, the model claims to reduce the average cost per task by over 60%, making it a more economical option within the Qwen lineup.
Pricing is set at $1.50 per million input tokens and $4.50 per million output tokens, with a reported task cost of $321. This makes it nearly three times cheaper than Qwen 3.8 Max and 11% less expensive than Kim K 3.1 Pro. These cost reductions are paired with improved capabilities in professional workflows, tool usage, and multimodal understanding.
Future additions to the Qwen 4 series are expected, including a 27 billion parameter open-weight model. Alibaba has announced that this smaller version is anticipated to launch within two to three weeks of Qwen 4 Max’s release.
Become an expert in Alibaba Qwen AI models with the help of our in-depth articles and helpful guides.
- Kimi K3.1, Qwen 4 and a Gemini 4.5 Argon? Nobody Expected THIS
- Alibaba Qwen 3.8 Rivals Claude Opus with Local 13.5GB RAM Operation
- What is Alibaba Qwen and its 6 LLM AI models?
Performance Highlights
Qwen 4 Max performs well in specific domains, particularly web development and front-end coding. On the Code Arena benchmark, it ranked sixth, demonstrating its ability to generate front-end designs. Tasks included creating a Mac OS clone with SVG-coded icons, animations, and functional apps, as well as implementing advanced visual features like glass morphism.
In game development, the model generated a playable 3D game using Unity, complete with visual effects, character animations, gameplay mechanics, and a refined HUD. It also created a Minecraft Dungeons clone featuring melee attacks, textures, and explorable areas across five floors. During testing, the host reached level two of the clone. However, these tasks often experienced significant latency; for instance, the Minecraft Dungeons clone took about two hours to generate.
The model further refined aspects of the 3D game, such as camera movement, lighting, animations, and gameplay. It also simulated GPU functionality, including tensor cores and TPUs, in a process that reportedly required six rounds of improvements and about a day to complete.
Professional Applications
Qwen 4 Max has shown potential in professional fields like finance, law, and healthcare. In finance, it reportedly used 90 MCP tools to conduct investment research and generate traceable Excel and PowerPoint reports. For legal tasks, it organized 198 documents and produced an 86-page due diligence report identifying 14 legal issues. In healthcare, the model analyzed approximately 1,400 medical images, leveraging its multimodal capabilities to generate patient recommendations that reportedly aligned with physicians’ decisions.
Benchmarks and Comparisons
Despite its strengths, Qwen 4 Max’s overall benchmark performance has been underwhelming. On the World of AI benchmark, it ranks behind leading models such as GPT 6 Sol, GPT 611 Terra, Opus 5, and Kim K 3.1 Pro. On Artificial Analysis’s intelligence index, it scored 43, a modest 9-point improvement over Qwen 3.8 Max but 12 points below GPT 6.1 Sol. Its Terminal Bench 4.0 score of 48% also falls significantly short of Kim K 3.1 Pro’s 67%.
However, on Alibaba’s internal benchmarks, Qwen 4 Max reportedly outperforms Opus 5.5, GPT 6.1 Sol, and Astron. It also surpasses Opus 5.5 on the Deep Suave 1.1 benchmark. On the Artificial Briefcase Knowledge Work benchmark, Qwen 4 Max holds a slight edge over Kim K 3.1 Pro.
Cost-effectiveness remains a challenge. While it uses 39% fewer output tokens than Qwen 3.8 Max, it costs more than twice as much per task as GPT 6.1 Sol, which also offers superior intelligence and faster response times. GPT 6.1 Sol provides better intelligence, is cheaper, and uses fewer tokens compared to Qwen 4 Max.
Strengths and Weaknesses
Qwen 4 Max excels in web development and multimodal understanding but struggles to compete with leading models in cost, efficiency, and overall practicality. Latency issues and slower response times further reduce its appeal, especially compared to faster, more token-efficient models like GPT 6.1 Sol.
The model’s intelligence per dollar value is another concern. While it shows improvements over its predecessor, it falls short of expectations for a flagship release, particularly given the extended wait for its debut. Early access tests reportedly yielded better results than the current version, adding to the sense of disappointment.
The host expressed frustration with the model’s overall performance, stating there is currently no incentive to use Qwen 4 Max due to the availability of better, cheaper, and more efficient alternatives. They had anticipated a more significant leap forward from Alibaba with this release.
Future Prospects
Alibaba plans to expand the Qwen 4 family with additional models, including a smaller open-weight version featuring 27 billion parameters, expected within the next two to three weeks. This release could address some of Qwen 4 Max’s shortcomings and provide a more competitive option for users seeking a balance between performance and cost.
Qwen 4 Max is a capable model with notable strengths in web development and professional tasks, but it falls short as a flagship release. Its high cost, slower performance, and underwhelming benchmark results make it less practical compared to leading competitors. For users focused on web development or specific professional applications, Qwen 4 Max offers valuable capabilities. However, for broader use cases, more efficient and cost-effective alternatives like GPT 6.1 Sol may be the better choice.
Media Credit: WorldofAI
Disclosure: Some of our articles include affiliate links. If you buy something through one of these links, Geeky Gadgets may earn an affiliate commission. Learn about our Disclosure Policy.