v139 · miscellaneous

Proofreader API

A JavaScript API for proofreading input text with suggested corrections, backed by an AI language model.

concepts

  1. Proofreader Probe

    On-device proofreader — finds grammar and spelling issues, suggests corrections. Runs against the same on-device model family as Summarizer / Translator / Prompt.

  2. Inline Corrections

    The structured corrections array, wired into a live editor. Marks errors in place, lists each correction with its type. Falls back to a rule-based proofreader so the page works without the on-device model.

  3. Proofreader vs Rewriter vs Summariser

    Same input, three sibling on-device APIs side by side. Word-level diff highlights the Proofreader's changes; Rewriter's output adapts to a tone selector; Summariser reports its compression ratio.

  4. Correction Review

    An accept / reject workflow: run the API on a document and step through each suggested correction individually. Accept applies the fix, reject skips it. A common editorial pattern now powered by on-device AI.

why it shipped

Browsers and operating systems are increasingly offering proofreading capability to help their users compose. By exposing this built-in model, we allow web applications to gain access to proofreading capability with the help of a language model, while avoiding every website needing to download their own multi-gigabyte language model, or sending input text to third-party APIs. The proofreader API in particular exposes a high-level API for interfacing with a language model in order to proofread inputs for a variety of use cases [1], with three specific proofreading functionalities: error correct

references