In successive version updates from v0.2.2 to v0.3.0, Qwen Code has introduced two core capabilities: Stream JSON support and a complete internationalization system. The new version provides streamed structured inputs and outputs through --output-format stream-json and --input-format stream-json, and works with a three-layer adapter architecture and session management mechanism to make it easier to embed SDKs, automation tools, and CI/CD pipelines to support continuous and structured drivers of agent workflows.
In terms of internationalization, Qwen Code has built-in interfaces in both Chinese and English, and allows for the expansion of more localized languages with custom language packs. Users can switch interface languages using /language ui zh-CN or specify model output languages individually through /language output Chinese, making it easy for developers around the world to flexibly switch between local habits and project needs. The official community is encouraged to contribute local language packs to improve multilingual support.
Security and stability are also important directions in this round of updates: the new version adds a 20MB buffer limit to mitigate the risk of memory exhaustion, fixes character issues caused by Windows encoding, enhances ripgrep binary detection and cross-platform compatibility, and refactors the authentication system to optimize authType management. In addition, ModelScope provider support and streaming option handling have been improved, integration tests have been fixed, and the hint completion logic and terminal notifications have been optimized, and a number of underlying fixes have significantly improved overall stability.
FAQs
Q: What is Qwen Code's Stream JSON feature?
A: It allows for interaction with models in streaming JSON form via --input-format/--output-format stream-json, facilitating structured communication in SDKs, automation scripts, and CI/CD.
Q: What does the three-layer adapter architecture and session management do?
A: Layer 3 adapters decouple protocols, models, and business logic, and session management ensures that multi-round conversations and multi-step tasks maintain contextual continuity in command-line and automation environments.
Q: What specific capabilities does internationalization support?
A: Built-in Chinese and English UI, you can switch between the interface language and the model output language through commands, and support developers to expand more local language packages.
Q: What are the key security and stability fixes?
A: Includes a 20MB buffer limit to prevent memory exhaustion, Windows encoding fixes, ripgrep detection enhancements, authentication system refactoring, and multiple fixes for integration testing and core logic.
Q: What are the typical use cases for these new features?
A: It is suitable for connecting Qwen Code to continuous integration pipelines, automated code review and refactoring tools, and command-line workflows for multilingual development teams.
