Qwen Code v0.10.6 新增模型與修復

🔑 增強重點摘要

• •Qwen Code is an open-source AI agent optimized for terminal use, adapted from Gemini CLI with enhanced parser and tool support specifically for Qwen-Coder models[3][4].
• •The Qwen3-Coder GitHub repository has garnered 18.1k stars, 1.6k forks, and contributions from 332 developers as of early 2026[4].
• •Qwen3-Coder supports a native 256K token context window, extendable to 1 million tokens, and uses long-horizon reinforcement learning for agentic coding[1].
• •Qwen-based coding models have exceeded 20 million global downloads, powering tools like Tongyi Lingma which has generated over 3 billion lines of code since June 2024[1].

🛠️ 技術深入

• •Qwen3-Coder achieves state-of-the-art performance among open-source models on SWE-Bench Verified benchmark without test-time or inference scaling, due to scaling in tokens, context length, synthetic data, and post-training with long-horizon reinforcement learning[1].
• •Qwen Code primarily uses TypeScript (91.2%), enabling composable, scriptable terminal operations like direct file editing, command execution, commits, and integration with external tools via MCP (e.g., Google Drive, Jira)[2][4].
• •New VS Code extension (Beta) provides a graphical IDE sidebar experience for Qwen Code, bypassing terminal requirements[2].

🔮 前景展望AI analysis grounded in cited sources

⏳ 時間線