Singapore – Alibaba Cloud, a digital and technology firm, has recently announced the company’s launch of more than 100 new large language models to the global open-source community. This initiative, called Qwen 2.5, was revealed during its annual flagship event, the ‘Apsara Conference.’
Apart from the initiative, the firm has also introduced a revamped full-stack infrastructure, aimed at addressing the increasing demands for robust AI computing.
This new infrastructure encompasses innovative cloud products and services that enhance computing, networking, and data centre architecture, designed to support the development and wide-range applications of AI models.
Eddie Wu, chairman and chief executive officer at Alibaba Cloud Intelligence, said, “Alibaba Cloud is investing, with unprecedented intensity, in the research and development of AI technology and the building of its global infrastructure. We aim to establish an AI infrastructure of the future to serve our global customers and unlock their business potential.”
As a newly launched LLM, the Qwen 2.5 release will see over 100 models being made open-source. Said range includes base models, instruct models, and quantised models of various precision levels and methods, ranging from different modalities like language, audio, and vision, along with specialised code and mathematical models.
The firm also shared that said models range from 0.5 to 72 billion parameters in size and are equipped with enhanced knowledge, stronger capabilities in math and coding, and are able to support over 29 languages. This approach makes it possible to cater to a wide array of AI applications both at the edge and in the cloud across various sectors, from automobiles to gaming to science research.
Jingren Zhou, chief technology officer at Alibaba Cloud Intelligence, also said, “Today marks a significant milestone as we launch our most expansive open-source initiative to date. This initiative is set to empower developers and corporations of all sizes, enhancing their ability to leverage AI technologies and further stimulating the growth of the open-source community.
“We remain committed to investing in advanced AI infrastructure to foster the widespread adoption of generative AI technologies across different industries,” he further emphasised.
In addition, Alibaba Cloud has also announced an update to its proprietary flagship model Qwen-Max, noting its performance on par with other state-of-the-art models in areas such as language comprehension and reasoning, math, and coding.
Interestingly, they have also unveiled a new text-to-video model as part of its image generator, the Tongyi Wanxiang large model family.
According to them, the new model is capable of generating high-quality videos in a wide variety of visual styles, from realistic scenes to 3D animation. In particular, it can generate videos based on Chinese and English text instruction and transform static images into dynamic videos.
Another significant deployment is an upgrade on its vision language model with the launch of Qwen2-VL, capable of comprehending videos lasting over 20 minutes and supporting video-based question-answering. The recently released Qwen2-VL is designed for integration into mobile phones, automobiles, and robots, facilitating the automation of specific operations.
In terms of computer programming, Alibaba Cloud has further launched an AI Developer, a Qwen-powered AI assistant designed to support programmers in automating tasks such as requirement analysis, code programming, and identifying and fixing software bugs.
Lastly, the firm has also announced substantial updates to its full-stack AI infrastructure, covering green datacenter architecture, data management, model training, and inferencing. This includes (1) a next-gen data centre architecture for surging AI development; (2) an open lake solution to maximise data utility; (3) an AI scheduler with integrated model training and inference; (4) a DMS for unified management of metadata; and (5) a more powerful elastic compute service.
These new updates aim to offer comprehensive support for customers and partners to maximise the benefits of the latest technology to build even more efficient, sustainable, and inclusive AI applications.
