Alibaba Cloud has significantly expanded its AI portfolio, announcing new foundational models, robust platform upgrades, and cutting-edge SaaS tools at its Spring Launch 2025. These initiatives underscore the company’s global ambitions to accelerate the adoption of artificial intelligence across diverse industries.
Scaling AI with the Qwen Model Suite
Central to the announcement is the international rollout of Alibaba Cloud’s proprietary Qwen large language model (LLM) series, now accessible via its Singapore availability zones. The lineup includes:
- Qwen-Max: A large-scale Mixture of Experts (MoE) model built for high-performance AI workloads.
- QwQ-Plus: A powerful reasoning model capable of tackling complex analytical and mathematical tasks.
- QVQ-Max: A visual reasoning model designed to process multimodal inputs and deliver chain-of-thought outputs.
- Qwen2.5-Omni-7b: A versatile end-to-end multimodal model for broad application scenarios.
These models provide global enterprises with powerful tools to build next-generation AI applications.
Platform Enhancements Drive AI Scalability
To support these advanced models, Alibaba Cloud has upgraded its Platform for AI (PAI) with scalable, cost-efficient capabilities.
Highlights include:
- Distributed inference via PAI-Elastic Algorithm Service (EAS), addressing the computational intensity of super-large models and ultra-long-text processing.
- Prefill-decode disaggregation, enhancing performance and cutting operational costs.
- Concurrency boost: A 92% increase in concurrent processing for the Qwen2.5-72B model.
- Token throughput: A 91% increase in tokens per second.
The PAI-Model Gallery has also been refreshed, now offering nearly 300 open-source models—including all of Alibaba Cloud’s Qwen and Wan series—through a no-code interface. New features such as model evaluation and distillation help enterprises efficiently scale AI deployment.
Integrating AI into Data Management
Alibaba Cloud’s cloud-native relational database PolarDB now features built-in AI inference powered by Qwen models. This enables:
- On-database inference with minimal latency
- Development of conversational RAG (Retrieval-Augmented Generation) agents
- Text embedding creation and semantic search capabilities
Meanwhile, AnalyticDB, Alibaba’s data warehouse, is now integrated into the Model Studio platform. Positioned as the recommended vector database for RAG-based generative AI solutions, this integration allows enterprises to connect proprietary knowledge bases and build intelligent, context-aware applications.
AI-Powered SaaS Tools for Digital Innovation
Alibaba Cloud also introduced two new SaaS tools to drive AI adoption in practical business contexts:
- AI Doc: A document processing tool that uses LLMs to parse forms, manuals, and reports. It can extract key information and generate tailored outputs, such as automated ESG reports when paired with Alibaba’s Energy Expert.
- Smart Studio: A generative AI platform for content creation across text-to-image, image-to-image, and text-to-video. It’s aimed at enhancing creative workflows in sectors like e-commerce, gaming, and entertainment.
Backed by Massive Investment
These advancements follow Alibaba’s February announcement of a $53 billion investment over the next three years in AI and cloud computing—more than its total spending in the previous decade.
“As cloud and AI become essential for global growth, we are committed to enhancing our core product offerings to address our customers’ evolving needs,” said Selina Yuan, President of International Business at Alibaba Cloud Intelligence.
With groundbreaking innovations, enterprise-grade infrastructure, and a bold investment strategy, Alibaba Cloud is positioning itself as a dominant force in the global AI and cloud computing landscape.
