Alibaba Cloud gathers the four top open-source models to further promote the explosion of AI applications

On February 25th, Alibaba Cloud Bianlian launched a Coding Plan (Development Package) that includes API services for four top open-source models: Qwen3.5, GLM-5, MiniMax M2.5, and Kimi K2.5. After subscribing to the package, users are no longer limited to a single model and can freely switch between multiple models, enjoying more stable and higher token quota services. According to reports, Alibaba Cloud is currently the only cloud provider worldwide offering this service.

The four models launched are all recent top-tier open-source models that have attracted significant attention in the AI community. They excel in programming and agent capabilities, providing powerful brains for AI tools like OpenClaw, enhancing agent productivity, and potentially driving the emergence of more diverse AI agent applications.

Qwen3.5 has set new efficiency and cost-performance benchmarks for large models, with a total of 397 billion parameters but only 17 billion activated. It performs excellently across reasoning, programming, and agent benchmarks. GLM-5 is the latest flagship open-source model from Zhipu, and its API service is in high demand since launch. MiniMax M2.5 is a model specifically designed for agent scenarios, performing well in advanced Excel processing, in-depth research, PowerPoint, and other Office applications. Kimi K2.5 has also achieved significant improvements in programming and agent capabilities.

Industry forecasts predict that China’s AI agent market will surpass 3.3 trillion yuan by 2028, with agents becoming widely adopted by enterprises and individuals. However, the explosion of agent applications not only requires powerful models but also solutions to the rising costs associated with their use. Agents perform complex tasks, involve multi-round planning and tool calls, and their token consumption grows exponentially. Data shows that the computing power required for a single agent can be 100 to 1,000 times that of traditional chatbots.

In this context, Alibaba Cloud has developed a new business model for models. The four major models include the Qwen large model, developed in-house by Alibaba, and the other three—Zhipu, Minimax, and Kimi—are backed by Alibaba Group investments. This combination better facilitates the deployment of these models.

Guo Tao, Deputy Director of the China Electronic Commerce Expert Service Center, told Securities Daily that this model leverages industry-specific integrated services to strengthen Alibaba Cloud’s core competitiveness in cloud services and AI, creating a differentiated advantage and effectively capturing market share. By aggregating top open-source AI models, the platform can continuously attract developers and ecosystem partners, rapidly improve the AI application ecosystem, and enhance user engagement and business expansion opportunities. This also demonstrates Alibaba Cloud’s technological integration and ecosystem operation capabilities in AI, further solidifying its position as a leading industry brand.

Notably, Alibaba Cloud is the first cloud provider to offer all four top open-source models and provides more cost-effective subscription packages, allowing users to enjoy top-tier model services. For example, the Alibaba Cloud Coding Plan Lite basic package offers up to 18,000 requests per month, with new users paying only 7.9 yuan for the first month. The Pro advanced package offers 90,000 requests per month, capable of handling thousands of complex tasks, with the first month costing only 39.9 yuan, significantly reducing costs for high-frequency coding scenarios.

“When top-tier model capabilities are accessible to more developers at low cost and high flexibility, the barriers to AI application innovation will decrease, potentially accelerating the transition from concept to implementation,” said Hong Yong, Deputy Researcher at the Ministry of Commerce Research Institute, to Securities Daily. He added that this service directly addresses the current surge in computing power consumption for agent applications, reducing developers’ trial-and-error and operational costs. Additionally, the flexible multi-model switching service allows developers to select the most suitable model for different scenarios such as programming, office work, and reasoning, improving development efficiency and quality.

Overall, Alibaba Cloud, as a leading cloud provider, possesses robust AI infrastructure, leading in chips, cloud infrastructure, and MaaS services, with close collaboration across various business units to meet diverse model demands.

Market research firm Omdia reports that in the first half of 2025, China’s overall AI cloud market will reach 22.3 billion yuan, with Alibaba Cloud accounting for 35.8%, surpassing the combined share of the second to fourth players. By 2025, Alibaba Cloud’s market share in China’s cloud market is expected to rise to 36%, further strengthening its leading position.

Guo Tao believes that Alibaba Cloud’s integrated multi-model service provides developers and small to medium-sized startups with a one-stop AI development capability, satisfying diverse model needs on a single platform and greatly simplifying integration processes while boosting R&D efficiency. With Alibaba Cloud’s underlying technology support, the service offers greater stability and technical assurance, allowing enterprises to avoid technical integration and service interruption risks and focus on core business innovation.