Alibaba Cloud, the digital technology and intelligence backbone of Alibaba Group, unveils new AI models, tools and infrastructure upgrades for its international customers, underscoring its ongoing commitment to driving AI innovation worldwide.
“We are launching a series of Platform-as-a-Service（PaaS) and AI capability updates to meet the growing demand for digital transformation from across the globe. These upgrades allow us to deliver even more secure and high-performance services that empower businesses to scale and innovate in an AI-driven world,” said Selina Yuan, President of International Business, Alibaba Cloud Intelligence.
Alibaba Cloud announced new offerings to international customers by expanding access to its foundational models and upgrading infrastructure products. Available through the company’s availability zones in Singapore, these models include the latest from its proprietary large language model (LLM) series, Qwen, such as the large-scale Mixture of Experts (MoE) model Qwen-Max, the reasoning model QwQ-Plus, the visual reasoning model QVQ-Max and the end-to-end multimodal model Qwen2.5-Omni-7b.
QwQ-Plus is an advanced reasoning model specialising in deep analytical thinking, tackling complex challenges like sophisticated QA tasks and expert-level math problems with precise, algorithm-driven solutions. Meanwhile, QVQ-Max is a visual reasoning model that effectively addresses complex multimodal problems with high accuracy and extended reasoning capabilities, supporting visual input and chain-of-thought output.
To further support the AI models on the PaaS front, Alibaba Cloud’s Platform for AI (PAI) has rolled out major enhancements to support scalable, cost-effective, and user-friendly solutions for generative AI and LLMs. PAI-Elastic Algorithm Service (EAS) debuts distributed inference capabilities with a multi-node architecture to satisfy the growing demands of super-large models fueled by the rise of MoE structure and ultra-long-text processing, addressing the limitations of traditional single-node architecture. To further boost performance and reduce costs, PAI-EAS introduces the prefill-decode disaggregation function which has led to a 92% increase in concurrency and a 91% boost in tokens per second (TPS) when deployed with the Qwen2.5-72B model, greatly improving scalability and efficiency.
PAI-Model Gallery has been upgraded to provide a comprehensive selection of nearly 300 cutting-edge open-source models, including the full range of Alibaba Cloud’s proprietary open-source models Qwen and Wan series, all accessible through a seamless, no-code deployment and management experience. It offers diverse deployment methods with underlying computing resources, along with new features like model evaluation for performance insights and model distillation, which reduces deployment costs by transferring knowledge from large to small models.
To enhance data management efficiency in the AI era, Alibaba Cloud has integrated its native AI inference capabilities – powered by Qwen – into its flagship cloud-native relational database PolarDB. With in-database machine learning capabilities, it eliminates data shifting typically required for inference workflow, significantly reducing processing latency while boosting efficiency and data security. Engineered for text-centric workloads, the new feature is ideal for scenarios including conversational RAG (Retrieval-Augmented Generation) agent development, text embedding generation, and semantic similarity search.
It also integrates its data warehouse AnalyticDB into Model Studio, Alibaba Cloud’s generative AI model and application development platform, as the recommended vector database for RAG solutions. This enhancement connects organizations’ proprietary knowledge bases directly to AI models and tools available on Model Studio, streamlining development of context-aware applications.
Alibaba Cloud also launched a new AI search function on its official website. Powered by Qwen, this AI assistant is designed to help potential enterprise clients, especially SMEs, speed up their solution discovery and gain key insights to facilitate strategic decision-making. It also offers access to cost-effective and scalable cloud solutions, along with free AI and cloud computing training resources.
Upgraded partner incentive policies were also introduced to better support resellers and distributors by providing increased flexibility, higher commission rates, and more rewarding opportunities for mutual growth. This commitment to empowering partners is further reinforced by enhanced training and support resources, designed to strengthen their capabilities and drive success within evolving ecosystem.
In February 2025, Alibaba Group announced an investment of US $53 billion (RMB 380 billion) over the next three years to advance its cloud computing and AI infrastructure, reinforcing its commitment to long-term technological innovation. This historic investment, which exceeds Alibaba’s total AI and cloud spending over the past decade, underscores the company’s ongoing dedication to AI-driven growth and its role as a leading global cloud provider. Alibaba Cloud now operates a global infrastructure network with 87 availability zones across 29 regions.
