Alibaba Cloud, the digital technology and intelligence backbone of Alibaba Group, today unveiled its latest large language model, Tongyi Qianwen. The technology innovator will integrate the new AI model into Alibaba’s various businesses to improve user experience in the near future. The company will also provide its customers and developers with access to the model to enable them to create customised AI features in a cost-effective way.
In addition, Alibaba Cloud announced lower cost options for key cloud products including their Elastic Compute Service (ECS) and Object Storage Service (OSS) by introducing new ECS instances, OSS Reserved Capacity (OSS-RC), and OSS Anywhere Reserved Capacity (OSS-ARC). The move aims to make computing more accessible and affordable for companies looking to unlock emerging opportunities in the new AI era.
“We have entered an unprecedented technological watershed moment, underpinned by generative AI and cloud computing, and businesses across all sectors have started to embrace intelligence transformation in order to stay ahead of their games,” said Daniel Zhang, Chairman and CEO of Alibaba Group and CEO of Alibaba Cloud Intelligence. “As one of the global leading cloud computing service providers, Alibaba Cloud is also committed to making computing and AI services more inclusive and accessible to enterprises and developers, so they can uncover more insights, explore new business models for expansion, and create more cutting-edge products and services for the society.”
Tongyi Qianwen will be integrated into all business applications across Alibaba’s ecosystem in the near future, from enterprise communication, intelligent voice assistance, e-commerce, search-to-navigation, and entertainment, to further enhance user experiences. Capable of answering questions in both Chinese and English languages, the model will be first deployed on DingTalk, Alibaba’s digital collaboration workplace and application development platform, and Tmall Genie, a provider of IoT-enabled smart home appliances.
- Tongyi Qianwen-empowered DingTalk is designed to make workplace communications more efficient. For example, it can summarize meeting notes, turn meeting conversations into texts, write emails, and create first drafts of business proposals or promotion campaign ideas using simple prompts. Users can also take a photo of a draft idea on paper and turn it into a mini application on DingTalk in a split second.
- Tongyi Qianwen-empowered Tmall Genie will be able to generate more engaging and vivid conversations with users. For instance, it can develop and tell stories for children, provide healthy diet recipes, offer travel tips, and recommend background music for a workout.
To further facilitate the intelligence transformation for enterprises to reap the benefits of AI-driven innovation, Alibaba Cloud will also offer its clients access to Tongyi Qianwen on the cloud and help them build their own customized large language models. By fine-tuning Tongyi Qianwen with customers’ proprietary intelligence and industrial know-how in a secure cloud environment, enterprises can establish their tailored AI models on the cloud to suit their specific business needs. This is expected to spark a new wave of growth momentum for customers, without them having to go through resource-intensive and expensive pre-training processes for building foundational models.
In addition, developers will soon be able to access Alibaba Cloud’s Tongyi Qianwen to create their AI applications at scale. This will further bolster the AI software ecosystem across sectors ranging from logistics to media, finance, manufacturing, energy, retail, and more. Multimodal capabilities including image understanding and text-to-image will be added to the Tongyi Qianwen model soon to provide users with more compelling AI features.
“Generative AI powered by large language models is ushering in an unprecedented new phase. In this latest AI era, we are equipped to create additional value for our customers and broader communities through our resilient public cloud infrastructure and proven AI capabilities,” said Jingren Zhou, CTO of Alibaba Cloud Intelligence. “We are witnessing a new paradigm of AI development where cloud and AI models are playing an essential role. By making this paradigm more inclusive, we hope to facilitate businesses from all industries with their intelligence transformation and, ultimately, help boost their business productivity, and expand their expertise and capabilities while unlocking more exciting opportunities through innovations.”
Tongyi Qianwen is based on Tongyi, Alibaba’s proprietary pre-trained model framework that unifies various AI models, including models that can turn text into images and short videos. Last year, Alibaba Cloud launched ModelScope, an open-source Model-as-a-Service (MaaS) platform that comes with hundreds of AI models including Tongyi-based text-to-image models for global developers and researchers. So far, with over 1 million active users, ModelScope has made 800 models available with over 16 million model downloads to date.
Designed for small and medium-sized (SME) enterprises, the new ECS Universal instance family provides the same stability as its similar product, while saving costs of up to 40%. It is suitable for SMEs conducting web applications and websites, enterprise office applications, and offline data analysis.
OSS-RC enables customers to reserve storage capacity in a specific region for a duration of one year. It reduces the capacity cost by up to 50% from pay-as-you-go prices. When customers have no requirement on storing their data in a specific region, they can create an OSS “Anywhere Bucket” to store data in a region chosen by Alibaba Cloud. OSS-ARC can then be used to reserve capacity for objects stored in OSS Anywhere Buckets. It reduces the capacity cost by up to 70% from pay-as-you-go prices.
In order to make computing resources more accessible to developers in China, Alibaba Cloud has also announced free trials of core products for up to three months, including ECS and PolarDB databases. Alibaba Cloud will also provide 1,000 free training courses on cloud technologies as well as around 500 hands-on experiments based on real business scenarios for developers to get easy access to cloud technologies.