Introduction:
Have programmers ever been tormented by repetitive code? Debugged complex logic into the middle of the night? Grown anxious over project documentation that seems never-ending? “Describe the requirement, and AI writes the code”—this was once a developer’s dream. The trend of the “coding vibe,” which swept through Silicon Valley, is now surging through the domestic market. First came ByteDance with the release of the country’s first AI IDE—Trae—and now Alibaba refuses to be left behind. In late May, it officially unveiled its new AI IDE: Tongyi Lingma.
1. Alibaba Cloud’s Tongyi Lingma: More Than a Plugin—A Full Standalone IDE Debuts
On May 30, Alibaba Cloud officially launched its first AI-native development environment tool—Tongyi Lingma AI IDE, bidding farewell to its earlier identity as a mere plugin and evolving into a fully functional integrated development environment. This leap forward means that developers no longer need to switch tools, but can enjoy an AI-empowered end-to-end coding experience on a unified platform. It also marks a significant step forward for Alibaba Cloud in the field of AI programming.
This is Alibaba Cloud’s first AI-native development tool, deeply integrated with the latest Tongyi Qianwen 3 large models. It also comes built-in with the largest domestic ModelScope MCP service platform, supporting one-click installation and deployment across more than 3,000 MCP service functions.
For example, with Gaode (Amap) MCP, users can generate a personalized travel guide web page in just 10 minutes—without writing a single line of code—and directly sync the finished itinerary to the Amap mobile app to create a customized travel route.
2. Alibaba Cloud Tongyi Lingma AI IDE: A Powerful New Tool
- Deep integration with Tongyi Qianwen 3 large models:
Tongyi Lingma AI IDE is deeply adapted to Alibaba Cloud’s latest Qianwen 3 models, which have achieved top scores among open-source models in benchmark datasets like LiveCodeBench and BigCodeBench, even surpassing some closed-source models in certain capabilities. This provides developers with more accurate smart suggestions and code completions, improving both quality and development efficiency. - Programming agent mode:
This introduces a programming agent model, where developers only need to describe a task. The agent can autonomously perform project awareness, code retrieval, terminal execution, MCP tool invocation, etc., truly completing end-to-end coding tasks. This dramatically saves developers’ time and energy, allowing them to focus more on creativity and logic design. - Auto-memory and inter-line suggestion prediction:
The IDE includes an automatic memory function that gradually forms a rich memory of developers’ habits, chat history, and project context, intelligently organizing it to improve collaboration. Its inter-line suggestion prediction function dynamically offers suggestions for the next code change based on current edits. Developers can press the Tab key to efficiently complete their coding—making the process smoother and more natural. - Extensive tool invocation capability:
Supporting the MCP protocol, it can call upon thousands of services from the ModelScope MCP marketplace—including dev tools, file systems, maps, etc.—greatly extending developers’ capabilities. For example, a developer can invoke Gaode MCP in Lingma to generate a personalized travel guide webpage and import the result into the Amap app to build a tailored travel map. - Inter-line chat feature:
Developers can interact with the IDE directly within the code editor area via natural language. This allows for in-file code modifications and real-time Q&A support from the AI IDE, significantly enhancing convenience and efficiency during development.
3. A Formidable Competitor: ByteDance’s Trae Moves Forward
While the industry is still digesting Alibaba’s new product, ByteDance showcased the rapid progress of Trae—Tongyi Lingma’s direct competitor—at its Volcano Engine FORCE conference.
In January this year, ByteDance took the lead in releasing the country’s first Chinese-language AI IDE—Trae. With functions like smart code completion, local code generation, and natural language programming, it provides domestic developers with highly efficient AI coding services. Since launch, Trae has been widely embraced by developer communities, and this latest update is particularly noteworthy.
- Over 1 million monthly active users, massive code output:
ByteDance VP of Technology Hong Dingkun announced that Trae now has over 1 million monthly active users. 80% of ByteDance’s engineers use it for daily development. It has already assisted developers in delivering more than 6 billion lines of code. - Feature-rich and practical:
Trae improves development efficiency through features like natural language-based project scaffolding, real-time code optimization, and multimodal code generation. Its Builder mode lets users generate full project code with just one description. The Chat mode assists in debugging and optimization.
Additionally, Trae supports graphics-to-code conversion—it can generate frontend pages from Figma designs or hand-drawn sketches. The newly released Cue (Context Understanding Engine) is similar to Alibaba’s inter-line prediction, capable of anticipating next steps for code modifications, generation, and error repair—streamlining the coding process. - Personalized support and commercial expansion:
In China, Trae integrates advanced models like Doubao 1.5-pro and DeepSeek to provide tailored support. Internationally, Trae launched a subscription model in May 2025—USD $3 for the first month, USD $10 per month thereafter.
This monetization strategy supports future development. The international version of Trae competes directly with Cursor and can access top global models like Claude Sonnet 4—but at half the price of its competitors, boosting its competitiveness in overseas markets.
Looking Ahead: Challenges Remain Despite the Revolution
Despite the revolutionary advantages shown by AI coding tools, several challenges remain:
- Capabilities in complex systems still need improvement. For areas such as microservice architecture and distributed transaction management—highly challenging from a design standpoint—AI’s current capabilities are still limited.
- Personalization needs further optimization. AI modules must better adapt to individual developers’ habits to enable deeply personalized collaboration.
- Code security and compliance must be ensured. It’s crucial to improve project-level copyright detection and vulnerability defense systems—particularly in commercial scenarios, where software asset security is paramount.
The competition in the AI programming field is heating up, and the future looks promising. The debut of Alibaba Cloud’s Tongyi Lingma AI IDE and ByteDance’s impressive Trae are only a snapshot of this fierce battle.
As AI technology continues to evolve, AI programming tools are becoming the standard in software development. They not only help developers improve efficiency and reduce errors but also make coding more enjoyable and creative. In the future, we can expect even more innovative products and features to emerge, further transforming the industry. Let’s wait and see what comes next.
[Disclaimer]: The above content reflects analysis of publicly available information, expert insights, and BCC research. It does not constitute investment advice. BCC is not responsible for any losses resulting from reliance on the views expressed herein. Investors should exercise caution.