Qwen MM-Plugins
About
Native multimodal plugins for Qwen models — make any AI agent harness multimodal-native. Ships 8 independent capability modules (Core/API/Search/Video-Memory/Video-Edit/Blender/FreeCAD/Edu-Agent), each installed as a skill + optional MCP server.
Our Verdict
Qwen MM-Plugins is an ambitious and well-structured multimodal plugin collection from the official QwenLM team. Its 8 modular capabilities (from local I/O to Blender 3D and FreeCAD) and single-installer approach make it a compelling choice for developers who want to add multimodal abilities to any compatible agent. The project is very young with limited stars and unfinished cookbooks, but the Apache 2.0 license, official team backing, and thoughtful architecture (skill + MCP server per module) suggest a promising trajectory. Worth trying if you work with Qwen models or need multimodal agent capabilities.
Pricing
Use Cases
- Adding multimodal file I/O (read images, video, docs) to any AI coding agent
- Using cloud APIs for vision understanding, OCR, ASR, and segmentation through an agent
- Driving Blender 3D modeling and FreeCAD parametric design from AI agent conversations
Pros
- 8 independent capability modules covering multimodal I/O, cloud APIs, search, video memory, 3D modeling, CAD and more
- Single guided installer supports all major harnesses (Claude Code, Codex, Qoder, OpenClaw, Qwen Code, Gemini CLI)
- Each module is a skill + optional MCP server — composable, only install what you need
- Apache 2.0 licensed, from the official QwenLM team
- Comes with cookbooks demonstrating each capability in action
Cons
- Project is very young (July 2026) — only 1,520 stars at verification time
- Windows support requires WSL2; native Windows not yet validated
- Some capabilities (API module) need third-party cloud API keys (e.g. DashScope)
- Several cookbooks are marked as TBD (not yet complete)
- Topics not set on GitHub repo — harder to discover organically
Installation
curl -fsSL https://raw.githubusercontent.com/QwenLM/Qwen-MM-Plugins/main/install.sh | bashFrequently Asked Questions
What is Qwen MM-Plugins?
It is an open-source multimodal plugin collection for Qwen models by Alibaba Cloud's QwenLM team, with 8 independent capability modules: Core (local I/O: read images/video/docs), API (cloud media understanding: VL/Omni A/V/ASR/segmentation), Search (web + reverse image search), Video-Memory (long-video hierarchical graph memory), Video-Edit (video editing and generation), Blender (3D modeling), FreeCAD (parametric CAD), and Edu-Agent (math/science tutorial video generation). Each module is installed as a skill + optional MCP server via a single guided installer script.
How do I install it?
One command: curl -fsSL https://raw.githubusercontent.com/QwenLM/Qwen-MM-Plugins/main/install.sh | bash. The installer handles all supported harnesses (Claude Code, Codex, Qoder, OpenClaw, Qwen Code, Gemini CLI) automatically. Windows users need WSL2. Alternatively, install manually: add the plugin marketplace, then install each capability module individually.
Is it free? What capabilities does it support?
Completely free and open source (Apache 2.0). 8 capability modules cover: local multimodal file I/O (Core), cloud API media understanding (VL/Omni A/V/ASR etc.), web + reverse image search (Search), long-video memory QA (Video-Memory), video editing and generation (Video-Edit), Blender 3D modeling and rendering (Blender), FreeCAD parametric CAD design (FreeCAD), and math/science tutorial video generation (Edu-Agent). Some capabilities (e.g. API module) require a cloud service API key like DashScope.