HeadlinesBriefing favicon HeadlinesBriefing.com

Chrome's New Prompt API Powers Browser AI

Hacker News •
×

Google has introduced the Prompt API for Chrome, enabling developers to integrate Gemini Nano directly in browsers. This API allows sending natural language requests for various applications like AI-powered search, personalized news feeds, content filtering, and calendar event creation. Developers can build Chrome Extensions that extract contact information or analyze web content automatically, transforming how users interact with online information.

The Prompt API requires specific hardware: Windows 10/11, macOS 13+, Linux, or ChromeOS on Chromebook Plus devices. Users need at least 22GB of free space, GPU with over 4GB VRAM, or CPU with 16GB RAM and 4+ cores. Developers can create sessions using the `create()` function, add context with initial prompts, and customize responses with parameters like temperature and top-K values, providing flexibility for different use cases.

Chrome's multimodal capabilities support text, image, and audio inputs across multiple languages. Developers can use audio transcription, image description for alt text, and create more intuitive user experiences. The API handles model downloads separately, and developers must acknowledge Google's Generative AI Prohibited Uses Policy before implementation, making it a comprehensive tool for browser-based AI applications that bridges the gap between web content and intelligent processing.