In April, Qoder CN released the coding agent, which can perform end-to-end coding tasks based on developer descriptions by using tools such as project retrieval, file editing, and terminal to modify multiple files within a project. At the same time, the AI Chat model service has been upgraded with the new Qwen3 series models.
Qoder CN adds Agent mode, which offers autonomous decision-making, environment awareness, and tool usage capabilities. It can perform end-to-end coding tasks based on developer coding needs by using tools such as project retrieval, file editing, and terminal. It also supports developer-configured MCP tools, making coding more aligned with the developer workflow.
At the same time, developers can switch between AI Chat, File Edit, and Agent modes within the same session flow — there is no need to create a new session to freely switch modes according to needs.
For details, see: Agent Mode.
The Qoder CN coding agent supports MCP tool usage. Based on user descriptions, the model autonomously plans and invokes MCP tools. It deeply integrates with the largest MCP Chinese community — ModelScope MCP Marketplace, covering more than 2400 MCP services across ten major popular domains including developer tools, file systems, search, and maps, comprehensively expanding the capability boundaries of the AI coding assistant and better aligning with the developer workflow.
For details, see: MCP.
Long-term memory capability has been added. During the developer's conversations with Qoder CN, rich memories about the developer, project, issues, and other topics gradually form, and related memories are automatically organized and updated. Memory capabilities help Qoder CN interact better with developers and, over time, understand them more deeply.
For details, see: Memory.
Qoder CN now fully supports Qwen3. Qwen3 adopts a Mixture-of-Experts (MoE) architecture with 235B total parameters and only 22B activated. Its parameter count is only 1/3 of DeepSeek-R1, dramatically reducing costs while surpassing top global models such as R1 and OpenAI-o1 in performance. Qwen3 is also China's first "hybrid reasoning model," integrating "fast thinking" and "slow thinking" into a single model — providing instant "quick" answers to simple requests with low compute, and multi-step "deep thinking" for complex questions, greatly saving compute consumption.
In AI Chat, context now supports project directories, allowing developers to ask questions or make modifications targeting a specific functional module. The context selection interaction has also been improved. For #file, #folder, #gitCommit, and #teamDocs, after selecting the corresponding type, you can search for and select specific content, and multi-select is supported for these content types.
For details, see: Context.
Qoder CN International is now live, supporting Alibaba Cloud International account login. Overseas developers can install Qoder CN in VS Code or JetBrains IDEs and log in directly using their Alibaba Cloud International account to start using Qoder CN.
Enterprise Dedicated Edition is now available on the International Site with a purchase flow. Enterprise customers can choose Singapore, Malaysia, or China (Shanghai) region for purchase.
Administrators of Enterprise Standard Edition and Enterprise Dedicated Edition can configure multiple inference model services and configure which model options are available to enterprise developers on the plugin side.
Qoder CN Coding Agent Released
Qoder CN adds Agent mode, which offers autonomous decision-making, environment awareness, and tool usage capabilities. It can perform end-to-end coding tasks based on developer coding needs by using tools such as project retrieval, file editing, and terminal. It also supports developer-configured MCP tools, making coding more aligned with the developer workflow.
At the same time, developers can switch between AI Chat, File Edit, and Agent modes within the same session flow — there is no need to create a new session to freely switch modes according to needs.
The File Edit mode is the former AI Developer module. It can be experienced in VS Code and JetBrains IDEs after upgrading the Qoder CN plugin to version 2.5.0 or above.
Supports MCP Tools and Integrates the MCP Marketplace
The Qoder CN coding agent supports MCP tool usage. Based on user descriptions, the model autonomously plans and invokes MCP tools. It deeply integrates with the largest MCP Chinese community — ModelScope MCP Marketplace, covering more than 2400 MCP services across ten major popular domains including developer tools, file systems, search, and maps, comprehensively expanding the capability boundaries of the AI coding assistant and better aligning with the developer workflow.
For details, see: MCP.
Added Memory Capabilities
Long-term memory capability has been added. During the developer's conversations with Qoder CN, rich memories about the developer, project, issues, and other topics gradually form, and related memories are automatically organized and updated. Memory capabilities help Qoder CN interact better with developers and, over time, understand them more deeply.
For details, see: Memory.
Added Support for Qwen3 Series Models
Qoder CN now fully supports Qwen3. Qwen3 adopts a Mixture-of-Experts (MoE) architecture with 235B total parameters and only 22B activated. Its parameter count is only 1/3 of DeepSeek-R1, dramatically reducing costs while surpassing top global models such as R1 and OpenAI-o1 in performance. Qwen3 is also China's first "hybrid reasoning model," integrating "fast thinking" and "slow thinking" into a single model — providing instant "quick" answers to simple requests with low compute, and multi-step "deep thinking" for complex questions, greatly saving compute consumption.
Context Supports Project Directories and Multiple File Selection
In AI Chat, context now supports project directories, allowing developers to ask questions or make modifications targeting a specific functional module. The context selection interaction has also been improved. For #file, #folder, #gitCommit, and #teamDocs, after selecting the corresponding type, you can search for and select specific content, and multi-select is supported for these content types.
For details, see: Context.
Personal Edition Supports Alibaba Cloud International Account Login
Qoder CN International is now live, supporting Alibaba Cloud International account login. Overseas developers can install Qoder CN in VS Code or JetBrains IDEs and log in directly using their Alibaba Cloud International account to start using Qoder CN.
Dedicated Edition Available on International Site, Supporting Regions such as Singapore and Malaysia
Enterprise Dedicated Edition is now available on the International Site with a purchase flow. Enterprise customers can choose Singapore, Malaysia, or China (Shanghai) region for purchase.
Enterprise Edition Supports Multi-Model Configuration
Administrators of Enterprise Standard Edition and Enterprise Dedicated Edition can configure multiple inference model services and configure which model options are available to enterprise developers on the plugin side.
Other Features and Improvements
- Supports navigation to files, methods, and definitions within the project in AI Chat;
- The AI Chat input box supports Shift+Enter for line breaks;
- When adding context in AI Chat, you can specify a knowledge base (Enterprise Edition);
- When adding a file as context in the chat window, related code files can be recommended based on the selected file;
- The @workspace project chat feature has been upgraded to #codebase, offering more flexible input combinations;
- Visual Studio side supports configuring Project Rules;
- Enhanced Go language effectiveness in inline code completion scenarios;
- Supports one-click issue reporting when startup fails;
- Improved network connection in firewall scenarios;
- Fixed some defects.