Agent-assisted development: one-stop Qwen3-VL on-device inference on Android
AI summary
A Tongyi Lab tutorial demonstrates how to use an Agent (Qoder) to help complete the full deployment of the Qwen3-VL-2B model on Android devices. The steps include: checking the JDK 21, NDK 27 and CMake 3.18.1 environments; creating an arm64-v8a Native C++ project (minSdk 29, compileSdk 35); downloading the roughly 1.4GB MNN/Qwen3-VL-2B-Instruct-MNN model via the ModelScope CLI; compiling libMNN.so with LLM vision support enabled (MNN_BUILD_LLM, LLM_SUPPORT_VISION); building the APK and pushing the model to the phone's private directory. The final app provides an image-text reasoning page that outputs the MNN version (v3.5.0), ABI and inference metrics. All the tedious steps can be executed automatically by the Agent.
Key Takeaways
- A Tongyi Lab tutorial demonstrates how to use an Agent (Qoder) to help complete the full…
- The steps include: checking the JDK 21,NDK 27 and CMake 3.
- 1 environments
- creating an arm64-v8a Native C++ project (minSdk 29,compileSdk 35)
- downloading the roughly 1.
Quick Facts
- Published
- Updated
- Source
- 公众号:通义实验室(千问)
- Category
- Tutorials & Tips
- AI score
- 68
用Agent辅助走通Qwen3-VL安卓端侧推理全流程,从环境搭建到JNI桥接一步到位。如果你在做移动端AI应用,这可能是目前最详细的实战指南,可以直接抄作业。
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: Agent-assisted development: one-stop Qwen3-VL on-device inference on Android. June 8, 2026. https://new.nat6.net/en/news/cmq51i5mc03ohslt2ggzp1vas/