Gemma 3n

Gemma 3n
Gemma 3n

原始链接: https://deepmind.google/models/gemma/gemma-3n/

专为响应式、低内存占用率的本地推理而设计，Gemma 3n 赋能新一代智能移动应用。 **功能：** * **多模态理解：**分析并响应图像和文本的组合——视频和音频支持即将推出。 * **隐私优先，离线可用：**构建注重用户隐私，并可在离线状态下可靠运行的智能交互式功能。 * **优化的设备性能：**移动优先架构，显著降低内存占用。由谷歌移动硬件团队和行业领导者共同设计。 * **动态资源使用：**4B 活动内存占用，并能够创建子模型以平衡质量和延迟。开始使用 Gemma 3n 进行开发

谷歌称Gemma 3是可在单GPU上运行的最强大的AI模型。 2025-03-20

Gemma 3 量化注意力模型：将 AI 带入消费级 GPU 2025-04-20

Gemma.cpp：用于 Gemma 模型的轻量级独立 C++ 推理引擎 2024-02-24

（评论） 2025-04-20

原文

Made for responsive, low-footprint local inference, Gemma 3n empowers a new wave of intelligent, on-the-go applications.

Capabilities

Multimodal understanding
Analyzes and responds to combined images and text — with video and audio coming soon.
Privacy-first, offline-ready
Build intelligent, interactive features that put user privacy first and work reliably offline.
Optimized on-device performance
Mobile-first architecture, with a significantly reduced memory footprint. Co-designed by Google's mobile hardware teams and industry leaders.
Dynamic resource usage
4B active memory footprint with ability to create submodels for quality-latency tradeoffs.

Gemma 3n Gemma 3n

Capabilities

Start building with Gemma 3n

Gemma 3n
Gemma 3n