In a groundbreaking announcement during the annual Apsara Conference, Alibaba Cloud has unveiled more than 100 open-source AI models under the Qwen branding, known as Qwen 2.5. This significant release highlights the company’s commitment to advancing AI technology in response to growing global demand.
Alibaba Cloud, the cloud computing arm of Alibaba Group, has also introduced an upgraded full-stack infrastructure designed to support robust AI computing. This new infrastructure features innovative products and services aimed at enhancing computing, networking, and data center architecture, thereby facilitating the development and diverse applications of AI models.
Eddie Wu, Chairman and CEO of Alibaba Cloud Intelligence, stated, “Alibaba Cloud is investing with unprecedented intensity in AI research and infrastructure development. Our goal is to establish a future-ready AI framework that empowers global customers to unlock their business potential.”
The newly launched Qwen 2.5 models, ranging from 0.5 to 72 billion parameters, showcase enhanced knowledge and improved capabilities in mathematics and coding. Supporting over 29 languages, these models cater to a variety of AI applications across sectors like automotive, gaming, and scientific research.
Since its initial release in April 2023, the Qwen model series has seen remarkable success, with over 40 million downloads across platforms such as Hugging Face and ModelScope, leading to the creation of over 50,000 derivative models on Hugging Face alone.
Jingren Zhou, CTO of Alibaba Cloud Intelligence, remarked, “This initiative is designed to empower developers and businesses of all sizes, enhancing their ability to leverage AI technologies and invigorating the open-source community.”
Alongside the open-source models, Alibaba Cloud announced an upgrade to its flagship model, Qwen-Max, which is reported to perform on par with other leading models in language comprehension, reasoning, and coding.
Expanding its multimodal capabilities, the company introduced a new text-to-video model as part of its Tongyi Wanxiang large model family, capable of generating high-quality videos in various styles from text instructions in both Chinese and English. Additionally, Alibaba Cloud launched Qwen2-VL, a vision language model that can analyze videos longer than 20 minutes and support video-based question-answering. The new AI Developer tool, powered by Qwen, assists programmers with tasks such as requirement analysis and code debugging.
To bolster these advancements, Alibaba Cloud is implementing several infrastructure upgrades, including:
- CUBE DC 5.0: A next-generation data center architecture that boosts energy and operational efficiency.
- Alibaba Cloud Open Lake: A solution designed to maximize data utility for generative AI applications.
- PAI AI Scheduler: A cloud-native scheduling engine for optimized resource management.
- DMS: OneMeta+OneOps: A platform for unified metadata management across multiple cloud environments.
- 9th Generation Enterprise Elastic Compute Service (ECS): Enhancements for improved application performance.
These developments position Alibaba Cloud as a key player in the AI landscape, equipping customers and partners with the tools necessary to build efficient, sustainable, and inclusive AI applications.
For more insights on AI and big data from industry leaders, check out the upcoming AI & Big Data Expo in Amsterdam, California, and London, co-located with other major events like the Intelligent Automation Conference and Cyber Security & Cloud Expo.
Sources:https://www.artificialintelligence-news.com/news/alibaba-cloud-unleashes-over-100-open-source-ai-models/,https://www.nexusgroup.com/partner/alibaba-cloud/
