Alibaba unveils Qwen3.8 AI model with 2.4 trillion parameters, claims second only to Fable 5

Alibaba unveils Qwen3.8, 'second only to Fable 5'

Days earlier, Moonshot AI set the record for the largest open-weight AI model yet. Alibaba shares surged more than 3.5pc in overnight trading on Monday (20 July) after the Chinese giant unveiled Qwen3.8-Max in preview - supposedly "second only to [Anthropic's] Fable 5". With 2.4trn parameters, this latest AI model in Alibaba's flagship series Owen is expected to go open-weight soon. It is available to try out on Alibaba's Token Plan, Qoder and QoderWork. Days earlier, Alibaba's Chinese contemporary Moonshot AI made a splash unveiling Kimi K3 which set the record for the largest open-weight AI model yet with 2.8trn parameters. Interest in the model surged much higher than expected following its launch, as new users swarmed to try the model forcing Moonshot to pause on taking new subscriptions on Sunday (19 July). Across cited benchmarks, K3 trailed behind only OpenAI's new GPT-5.6 Sol and Anthropic's Fable 5, while outperforming them in certain coding and general agent tasks at a cheaper price. The latest launches from Chinese AI companies have sent shares of rival AI model maker Zhipu down by more than 45pc in the past five days. MiniMax, which went public earlier this year shortly following Zhipu, has also lost nearly 13pc of its share value in the past five days and more than 61pc of its value over the past month. This, as Moonshot AI reportedly plans to go public as early as in six months, riding on the growing wave of interest in the cheaper Chinese AI models, despite efforts from Washington to curb semiconductor exports to the country. US curbs on semiconductor exports to China over recent years has seemed to have pushed China to focus on its own chip capabilities. Earlier this year, the country reportedly began restricting overseas travel for its top AI professionals to ensure its talent remains beneficial at home. Meanwhile, US lawmakers are urging the Trump administration to ban US companies from buying memory chips from Chinese semiconductor developers. Don't miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic's digest of need-to-know sci-tech news.

Alibaba previews Qwen3.8, claims it's second only to Claude Fable 5

Alibaba Group Holding Ltd. today previewed the most powerful artificial intelligence model in its Qwen family, Qwen3.8, and claimed the system trails only Anthropic PBC's Claude Fable 5 among the world's frontier models. The company revealed the model, offered as a preview build called Qwen3.8-Max-Preview, on Sunday at the World Artificial Intelligence Conference in Shanghai. It carries 2.4 trillion parameters, making it the first Qwen model above 1 trillion parameters to process images, video and documents alongside text. Alibaba's Qwen team said in a post on X that the model ranks "second only to Fable 5," a reference to Anthropic's recent release. It offered no benchmark scores to support the claim and no independent evaluation has verified the ranking. That absence stands out. Alibaba's previous flagship, Qwen3.7-Max, shipped in May with a full set of published results, including a score of 56.6 on the Artificial Analysis Intelligence Index. The Qwen3.8 preview arrived with no model card, no activated-parameter count and no benchmark data at all. Alibaba published no task-level comparison with Qwen3.7-Max and described the new model only as "continuously evolving." Open weights are promised "soon," though the company has not set a date or published license terms. The preview is available now through Alibaba's Token Plan subscription and its Qoder and QoderWork developer platforms, priced at 10% of the standard rate during the trial period. The timing places Qwen3.8 directly against a domestic rival. Chinese AI startup Beijing Moonshot AI Technology Co. Ltd. released Kimi K3, a 2.8 trillion-parameter model, three days earlier. Both launches push Chinese developers into the multi-trillion-parameter tier, a scale until recently associated mainly with the largest U.S. labs Alibaba has made Qwen central to its effort to position itself as China's default AI supplier. Over the past year the company has released a steady run of open-weight models while building out cloud infrastructure and custom chips to run them. The open-weight approach has helped Qwen assemble one of the largest developer followings of any Chinese model family, a base it now hopes to carry into the trillion-parameter class. For enterprise buyers weighing the announcement, the comparison with Fable 5 is Alibaba's alone. No independent leaderboard has scored Qwen3.8 yet, and the last Qwen model that was ranked, Qwen3.7-Max, sits well down LMArena's list, where Fable 5 is No. 1. Benchmarks and the promised open weights would settle it, but Alibaba has released neither.

Alibaba launches Qwen3.8-Max-Preview, says it trails only Anthropic's Fable 5

Alibaba has launched its new Qwen3.8-Max-Preview AI model with 2.4 trillion parameters. The preview version is available through Alibaba's Token Plan, Qoder, and QoderWork. To mark the launch, Alibaba is offering the model at 10% of the standard price. It is also offering an exclusive night-time rate with discounts of up to 98%. Alibaba has introduced Qwen3.8-Max-Preview, the latest addition to its Qwen family of foundation models, ahead of its planned open-weight release. "With a massive 2.4T parameters, this model is continuously evolving. We believe it's one of the most powerful model available today, compatible to leading frontier AI models, second only to Fable 5," the company said in an X post.The preview version is available through Alibaba's Token Plan, Qoder,

Qwen 3.8 Max vs Fable 5: is Alibaba's AI Closing the Gap?

Alibaba's Qwen 3.8 Max stands out with its 2.4 trillion parameters and a remarkable 1 million token context window, positioning it as a serious competitor in the AI space. Highlighted by World of AI, this model's ability to process text, images and video makes it particularly suited for large-scale, multimodal tasks. For instance, its performance in front-end development demonstrates precision in generating responsive web layouts and animated SVG designs, though its slower output speeds may pose challenges for time-sensitive projects. These strengths and trade-offs reflect its potential as a versatile yet evolving system. Explore how Qwen 3.8 Max stacks up against rivals like Fable 5 and Kim K3 in areas such as 3D game prototyping, where it excels in creating basic open-world environments but struggles with delivering polished, high-detail outputs. Gain insight into its multimodal capabilities, the trade-offs between accuracy and efficiency and the specific scenarios where it shines. This feature also previews Alibaba's plans for the Qwen 4.0 series, which aims to address current limitations and refine the model's performance for broader applications. Features and Multimodal Capabilities Qwen 3.8 Max represents a leap forward in AI technology, particularly in its ability to handle multimodal content. Its architecture is tailored to support a range of sophisticated tasks, including: * Generating interactive designs for web and application interfaces * Creating immersive 3D game environments * Maintaining coherence across extensive and complex inputs The 1 million token context window is a standout feature, allowing the model to process intricate, large-scale projects such as open-world simulations or detailed front-end development. This capability makes Qwen 3.8 Max a valuable tool for professionals working on high-volume, detail-oriented tasks. However, its performance in certain areas, such as speed and refinement, highlights the need for further optimization to meet the demands of time-sensitive or highly polished projects. Performance in Front-End Development and SVG Animation One of the most notable strengths of Qwen 3.8 Max lies in its application to front-end development. The model excels in generating: * Responsive and visually appealing web layouts * Detailed and dynamic SVG animations * Interactive landing pages with advanced functionality For example, Qwen 3.8 Max has successfully replicated Mac OS-style interfaces and created animated web designs featuring interactive elements. These capabilities make it a powerful tool for developers and designers seeking to streamline their workflows. However, the model's iterative refinement process, while making sure high precision, results in slower output generation compared to faster-performing competitors like Kim K3. This trade-off between accuracy and speed may limit its utility for projects requiring rapid turnaround times. Advance your skills in Qwen by reading more of our detailed content. Capabilities in 3D Game Development In the realm of 3D game creation, Qwen 3.8 Max demonstrates potential but falls short of delivering the level of detail and realism required for high-end gaming experiences. The model is capable of generating: * Basic open-world environments * Interactive elements for prototyping These features make it suitable for early-stage development and concept testing. However, when compared to competitors like Fable 5, Qwen 3.8 Max struggles to produce the polished, visually refined outputs necessary for fully realized gaming projects. This limitation underscores the need for further advancements in its ability to handle complex compositions and intricate visual details. Strengths and Weaknesses in Comparison When evaluated against leading AI models, Qwen 3.8 Max presents a mixed performance profile. Its strengths include: * High precision in tasks such as SVG animation and coding * Robust handling of large-scale, multimodal inputs However, the model also exhibits notable weaknesses: * Limited speed and efficiency in output generation * Challenges in delivering cohesive, polished results for complex tasks For instance, while Qwen 3.8 Max excels in generating detailed animations and responsive designs, models like Kim K3 outperform it in producing refined and cohesive outputs for intricate design projects. Additionally, the absence of public benchmarks and detailed architectural insights makes it difficult to fully assess Qwen 3.8 Max's capabilities relative to its competitors. This lack of transparency may hinder its adoption among developers and researchers seeking a comprehensive understanding of its performance. Future Developments and Prospects Alibaba has already announced plans to address the limitations of Qwen 3.8 Max with the upcoming Qwen 4.0 series. Key improvements are expected to include: * Faster processing speeds to enhance efficiency * Improved output quality for more polished results * Enhanced handling of complex, large-scale tasks These advancements aim to close the gap between Qwen 3.8 Max and top-tier models like Fable 5 and Kim K3. If successful, the Qwen series could become a stronger contender in areas such as interactive design, large-scale simulations and advanced game development. Alibaba's commitment to innovation and refinement suggests that future iterations of the Qwen series may address the current model's shortcomings while building on its strengths. Access and Availability At present, Qwen 3.8 Max is available through Alibaba's token-based plans, offering limited access for testing and experimentation. While API access has not yet been introduced, broader availability is anticipated as the model matures. This phased rollout strategy allows Alibaba to gather valuable user feedback and make iterative improvements before expanding the model's reach. For developers and organizations interested in exploring Qwen 3.8 Max's capabilities, this limited access provides an opportunity to evaluate its potential for specific applications. Looking Ahead Qwen 3.8 Max represents a promising addition to the AI landscape, offering notable strengths in front-end development, SVG animation and basic 3D game creation. However, its slower processing speeds and challenges in handling complex, polished tasks highlight areas for improvement. As Alibaba continues to refine the Qwen series, its ability to compete with leading models will depend on addressing these limitations and enhancing its performance across key domains. For now, Qwen 3.8 Max serves as a capable tool for specific use cases, with significant room for growth in the rapidly evolving AI ecosystem. 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.

Qwen 4.0 Leak Reveals Possible September 2026 Launch

Alibaba's Qwen 4.0, DeepSeek V4 GA and GLM 5.3 represent significant developments in the AI field, each addressing distinct challenges and opportunities. According to World of AI, Qwen 4.0 has reportedly been leaked ahead of schedule, revealing capabilities in complex 3D coding and design tasks that could support industries requiring advanced spatial modeling. DeepSeek V4 GA is also on the horizon, with its public launch slated for Monday, July 20, 2026, aiming to make advanced AI functionalities more widely accessible. Dive into the practical applications of Qwen's 3D design capabilities and how they align with industry needs. Understand the expanded reach of DeepSeek V4 GA and its potential impact on accessibility in AI. Learn about GLM 5.3's focus on scalability and how distillation techniques are being used to optimize performance while managing resource demands. Alibaba's Qwen Models: Redefining AI Capabilities Alibaba continues to solidify its position as a leader in AI innovation with its upcoming Qwen 3.8 and Qwen 4.0 models. Qwen 3.8, currently in the final stages of testing, is set for release in August 2026, with Qwen 4.0 following closely in September. These models are expected to outperform GLM 5.2, particularly in handling complex 3D coding and design tasks, marking a significant leap in AI capabilities. In addition to these models, two stealth projects, codenamed "Caleb" and "Terrania Alpha" -- have surfaced during testing. These experimental models exhibit remarkable proficiency in generating intricate 3D environments, positioning Qwen as a frontrunner in AI-driven design and coding. Their potential applications are vast, ranging from video game development to industrial design and architectural visualization. This versatility underscores Alibaba's commitment to pushing the boundaries of what AI can achieve. DeepSeek Version 4GA: Broadening AI Access DeepSeek's Version 4GA represents a significant milestone in the widespread access of AI technology. Currently undergoing a phased grayscale rollout, this model is gradually being introduced to a broader audience. Early feedback from testers highlights substantial improvements in coding accuracy, design precision and overall output quality, making it a promising tool for developers and creators. There is speculation within the industry that DeepSeek 4GA incorporates advanced distillation techniques derived from proprietary models like Fable 5. If confirmed, this would reflect a growing trend in the AI sector: using distillation to enhance model performance while maintaining cost efficiency. A full public release of DeepSeek Version 4GA is anticipated as early as Monday, July 20, 2026, potentially setting a new benchmark for accessible AI solutions. Check out more relevant guides from our extensive collection on Qwen that you might find useful. GLM 5.3: The Next Challenger GLM is preparing to unveil its next-generation model, GLM 5.3, with public testing already underway. Although specific details about its capabilities remain scarce, the model is expected to build upon the strengths of its predecessor, GLM 5.2 and directly challenge Alibaba's Qwen series. A release date is projected for late 2026, adding further momentum to the competitive landscape of AI development. Industry insiders suggest that GLM 5.3 will focus on enhancing scalability and adaptability, making it a versatile tool for various applications. This aligns with the broader industry trend of creating AI systems that are not only powerful but also flexible enough to meet diverse user needs. Performance Highlights of Emerging AI Models The latest advancements in AI are setting new benchmarks for performance and innovation. Key highlights include: * Qwen 3.8 and Qwen 4.0: These models excel in 3D coding and design, delivering outputs that are both precise and visually compelling. Their applications span industries such as gaming, architecture and virtual reality, showcasing their versatility and impact. * DeepSeek Version 4GA: Known for its adaptability, this model has demonstrated the ability to create hybrid applications, including a game that blends elements of Minecraft and No Man's Sky. This capability highlights the fantastic potential of AI in creative industries. * GLM 5.3: While details remain limited, early testing suggests that this model will prioritize scalability and adaptability, making it a strong contender in the competitive AI market. Industry Trends: Distillation and Cost Efficiency As competition within the AI sector intensifies, developers are increasingly focusing on delivering high-quality models while optimizing costs. Distillation practices, where smaller models are trained using outputs from larger proprietary systems, are becoming a common strategy. This approach not only accelerates the development process but also reduces the computational resources required, making advanced AI technology more accessible. However, the use of distillation raises ethical questions, particularly regarding intellectual property and transparency. Despite these concerns, the practice is gaining traction as a means to balance innovation with cost efficiency. Leading AI labs are also exploring other methods to optimize resources, making sure that innovative technology can reach a wider audience without compromising on quality. The Road Ahead for AI Innovation The pace of AI development shows no signs of slowing, with companies like Alibaba, DeepSeek and GLM leading the charge. From the new 3D coding capabilities of Qwen 4.0 to the hybrid applications enabled by DeepSeek Version 4GA, these advancements are reshaping the possibilities of AI across industries. As competition drives innovation, the focus remains on improving performance, accessibility and cost efficiency. For developers, businesses and end-users, the future of AI holds immense potential to transform how we work, create and interact with technology. 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.

Alibaba Says New AI Model Is Just Second to Anthropic's Fable 5

Alibaba Group previewed its new artificial-intelligence model that it says is second only to Anthropic's Fable 5, the latest sign that China is fast closing the gap with the U.S. for global technological dominance. The Qwen3.8-Max-Preview, the preview version of the company's latest flagship large language model, boasts 2.4 trillion parameters, and is available on Alibaba's coding platforms, the Hangzhou-based company said in a post on X. Alibaba describes the new model as among the most powerful available today, comparable to leading frontier AI models and second only to Fable 5, Anthropic's flagship Mythos-class model. It plans to make the model open-weight soon, meaning it will be available for developers to download and modify. The Sunday preview release came days after China's Moonshot launched its Kimi K3 model, surprising many in Silicon Valley with its advanced capabilities. The Kimi K3 model has 2.8 trillion parameters dictating its decision-making and is the world's biggest open-source model, according to the Beijing-based AI startup. Kimi K3's performance surpassed another recent model launched by Beijing-based Z.AI, the GLM-5.2, across major benchmarks. AI system parameters work like brain cells: The more a model has, the more knowledge it can store, making the count a shorthand for a model's capabilities. On Monday, Alibaba's Hong Kong-listed stock rose as much as 5.6% before paring gains to 5.15% at midday, outperforming the Nasdaq-like Hang Seng Tech Index's 2.9% increase. Shares of Z.AI, previously known as Zhipu AI, and MiniMax fell a further 13% and 5.7%, respectively, following the powerful model releases by rivals. Analysts say that leadership in the large language model space is changing quickly. "Competition is likely to intensify, eventually leaving only a small group of large language model developers capable of funding frontier-model development," said Laila Khawaja, research director at Gavekal Technologies. Write to Tracy Qu at [email protected] -0- The recent streak of powerful model releases buoy views that Chinese AI labs are closing in on U.S. rivals at an accelerating pace.