Alibaba Cloud opens up a multi-model coding hub for developers

Alibaba Cloud (阿里云) has made a move that could seriously reshape the landscape of AI coding tools in China. The company has updated its Coding Plan subscription service to support several competing models at once: its own Qwen 3.5-Plus (千问3.5), plus GLM-4.7 from Zhipu AI (智谱) and Kimi-K2.5 from Moonshot AI (月之暗面). With a single subscription, developers can now switch freely between models and pick the best tool for each task.

Another AI subscription launch on its own would surprise few people — the market is already flooded with offerings. What matters here is how the product is built. Alibaba Cloud isn't just selling access to its own model; it's building an aggregator platform where models from direct competitors sit on the same shelf as its flagship Qwen. That's a fundamentally different approach from the one most vendors take: typically, each company pushes only its own model and locks users into its own ecosystem.

On the technical side, Coding Plan integrates with the key AI coding tools: Qwen Code, Claude Code, Cline, and OpenClaw. Developers can keep working in a familiar environment while the subscription acts as a single backend, serving different models through a standardized interface. The inclusion of Claude Code among the supported clients is notable: it shows that Alibaba isn't trying to wall users off from Western tools. Instead, it wants to become a universal inference provider.

The model lineup isn't random either. Qwen 3.5-Plus is the latest iteration of Alibaba's flagship model, and it posts strong code generation and analysis results in recent benchmarks. GLM-4.7, from Beijing-based Zhipu AI (a company with close ties to Tsinghua University), is known for deep context understanding and for handling long stretches of code. Kimi-K2.5, from startup Moonshot AI, has become hugely popular in China thanks to its ability to process ultra-long contexts, which is critical when working with large codebases. Together, the three models cover different use cases, and the ability to switch between them at no extra cost delivers real value to professional developers.

A bigger strategy sits behind this decision. Alibaba Cloud, Asia's largest cloud provider, is clearly betting on the infrastructure layer rather than the model layer. The logic is simple: models are becoming increasingly interchangeable, and benchmark leadership changes hands every few months, but the infrastructure through which those models are consumed locks in long-term customer relationships. It's the same logic Amazon Web Services applies with Bedrock and Microsoft with Azure AI Studio, both of which offer access to models from multiple vendors through a single platform.

For China's AI coding market, this is an important signal. Until now, competition between Alibaba, Baidu, ByteDance, Zhipu AI, and Moonshot AI has played out mostly at the model level, with each company trying to prove that its model writes better code than the rest. Alibaba Cloud's move shifts the focus: the company is effectively saying that the race for the best model matters less than controlling the distribution channel. If developers get used to accessing every model through Coding Plan, Alibaba Cloud will become the intermediary for a significant share of AI coding in China — regardless of whose model happens to be on top at the moment.

Pricing and access terms remain an open question: Alibaba Cloud has not yet disclosed details of the updated plans. But simply bundling competing models under one subscription sets a new standard for the market. Other Chinese cloud providers — Huawei Cloud, Tencent Cloud, Baidu AI Cloud — will either have to follow suit or make a convincing case for why being tied to a single model beats freedom of choice. At a time when no model holds the lead for more than a few months, the case for a multi-model approach looks increasingly compelling.