AI Without a Backend: The Browser's Built-in AI APIs for Web Developers

Short summary

Chrome now ships built-in AI APIs powered by Gemini Nano, enabling web developers to run language models directly in the browser with no API keys, backend proxies, or per-token billing. The APIs include task-specific options (Summarizer, Translator, Writer, Rewriter, Language Detector) and a multimodal Prompt API, all sharing a common feature-detect, availability-check, create-and-use lifecycle. While currently desktop-only with a ~4GB initial model download, these APIs are production-viable as progressive enhancements for use cases like review summarization and image-to-structured-JSON conversion.

• •Chrome ships on-device AI via Gemini Nano — no API key, no backend, no per-token cost
• •Task APIs (Summarizer, Translator, Writer, etc.) plus a general-purpose multimodal Prompt API
• •Desktop-only with ~4GB first-download; best used as progressive enhancement with graceful degradation

Generated with AI, which can make mistakes.