v137 · miscellaneous

Rewriter API

The Rewriter API transforms and rephrases input text in requested ways, backed by an on-device AI language model. Developers may use this API to remove redundancies within a text in order to fit into a word limit, rephrase messages to suit the intended audience or to be more constructive if a message is found to use toxic language, rephrasing a post or artic

concepts

  1. Rewrite

    On-device Rewriter — refactor text by tone/length/format without going to a server. Same model family as Summarizer / Translator.

  2. Kindness pass

    One-click pass that takes a sharp draft and rewrites it more kindly — useful for code-review comments, support replies, and PR feedback.

  3. Rewrite lab

    Run four rewrites in parallel (tone × length grid), word-diff each against the original, and surface timings, word counts, and reading time so the trade-offs are visible.

  4. Streaming rewriter

    Live editor that re-streams on every dial twist with abort signalling. Meters track total streams, cancelled streams, and average first-token latency — the cost of an interactive UI on top of a streaming model.

why it shipped

Browsers and operating systems are increasingly expected to gain access to a language model. By exposing this built-in model, we avoid every website needing to download their own multi-gigabyte language model, or send input text to third-party APIs. The rewriter API in particular exposes a high-level API for interfacing with a language model in order to transform inputs for a variety of use cases [1], in a way that does not depend on the specific language model in question.

references