 Thomas Steiner [GitHub](https://github.com/tomayac) [LinkedIn](https://www.linkedin.com/in/thomassteinerlinkedin) [Mastodon](https://toot.cafe/@tomayac) [Bluesky](https://bsky.app/profile/tomayac.com) [Homepage](https://blog.tomayac.com/)  Alexandra Klepper [GitHub](https://github.com/alexandrascript) [Bluesky](https://bsky.app/profile/alexandrascript.com)
Published: September 12, 2025
| Explainer | Web | Extensions | Chrome Status | Intent |
|---|---|---|---|---|
| [GitHub](https://github.com/explainers-by-googlers/proofreader-api) |  Developer trial |  Developer trial | [View](https://chromestatus.com/feature/5164677291835392) | [Intent to Experiment](https://groups.google.com/a/chromium.org/g/blink-dev/c/Gboyyec4qmg/m/_l6n39OKBAAJ) |
Proofreading is the process of looking for and correcting errors in grammar,
spelling, and punctuation. Browsers and operating systems have increasingly
offered proofreading to their composing tools, such as in
[Google Docs](https://support.google.com/docs/answer/57859).
With the Proofreader API, you can provide interactive proofreading to your web
application or extension, with built-in AI. This API offers the following
functions:
- **Correction**: Correct user inputs for grammar, spelling, and punctuation.
- **Labels**: Label each correction by the error type.
- **Explanation**: Defining what the error is or why the correct was necessary in plain language.
> [!IMPORTANT]
> **Important** : The built-in foundation model is a generative AI model. Before you build with APIs that use it, you should review the [People + AI Guidebook](https://pair.withgoogle.com/guidebook/) for best practices, methods, and examples for designing with AI.
### Use cases
There are many reasons you may want to use Proofreader API For example:
- Suggest corrections to forum messages, comments on articles, and emails, before the post is submitted.
- Provide corrections during active note-taking.
Is your use case missing? Join the [early preview program](https://developer.chrome.com/docs/ai/join-epp.md.txt)
to share your feedback.
## Get started
Join the Proofreader API origin trial, running in Chrome 141 to 145.
### Review the hardware requirements
The following requirements exist for developers and the users who operate features using these
APIs in Chrome. Other browsers may have different operating requirements.
The **Language Detector** and **Translator APIs** work in Chrome on
desktop. These APIs do not work on mobile devices.
The **Prompt API** , **Summarizer API** , **Writer API** ,
**Rewriter API** , and **Proofreader API** work in Chrome when the
following conditions are met:
- **Operating system** : Windows 10 or 11; macOS 13+ (Ventura and onwards); Linux; or ChromeOS (from Platform 16389.0.0 and onwards) on [Chromebook Plus](https://www.google.com/chromebook/chromebookplus/) devices. Chrome for Android, iOS, and ChromeOS on non-Chromebook Plus devices are not yet supported by the APIs which use foundation models.
- **Storage** : At least 22 GB of free space on the volume that contains your Chrome profile.
> [!NOTE]
> Built-in models should be significantly smaller. The exact size may vary slightly with updates.
- **GPU or CPU** : Built-in models can run with GPU or CPU.
- **GPU**: Strictly more than 4 GB of VRAM.
- **CPU**: 16 GB of RAM or more and 4 CPU cores or more.
- **Note**: The Prompt API with audio input requires a GPU.
- **Network** : Unlimited data or an unmetered connection.
> [!IMPORTANT]
> **Key term** : A [metered connection](https://support.microsoft.com/windows/metered-connections-in-windows-7b33928f-a144-b265-97b6-f2e95a87c408) is a data-limited internet connection. Wi-Fi and ethernet connections tend to be unmetered by default, while cellular connections are often metered.
> [!NOTE]
> **Note**: The network requirement is only for the initial download of the model. Subsequent use of the model does not require a network connection. No data is sent to Google or any third party when using the model.
Gemini Nano's exact size may vary as the browser updates the model. To determine the current size, visit `chrome://on-device-internals`.
> [!NOTE]
> **Note**: If the available storage space falls to less than 10 GB after the download, the model is removed from your device. The model redownloads once the requirements are met.
### Add support to localhost
All of the built-in AI APIs are available on `localhost` in Chrome. Set the
following flag to **Enabled**:
- `chrome://flags/#proofreader-api`
Then click **Relaunch** or restart Chrome. If you encounter errors,
[troubleshoot localhost](https://developer.chrome.com/docs/ai/get-started.md.txt#troubleshoot_localhost).
### Sign up for the origin trial
To start using the Proofreader API, follow these steps:
1. Go to the [Proofreader API origin trial](https://developer.chrome.com/origintrials/#/registration/2794008579760193537).
2. Click **Register** and fill out the form. In the Web origin field, provide your [origin](https://web.dev/articles/same-site-same-origin#origin) or extension ID, `chrome-extension://YOUR_EXTENSION_ID`.
3. To submit, click **Register**.
4. Copy the token provided, and add it to every participating web page on your origin or include it in your Extension manifest.
- If you're building an Extension, follow the [Extensions origin trial instructions](https://developer.chrome.com/docs/web-platform/origin-trials.md.txt#extensions)
5. Start using the Proofreader API.
Learn more about how to [get started with origin trials](https://developer.chrome.com/docs/web-platform/origin-trials.md.txt).
## Use the Proofreader API
To determine if the model is ready to use, call
[Proofreader.availability()](https://developer.chrome.com/docs/ai/get-started.md.txt#model_download). If the
response to `availability()` was `"downloadable"`, listen for download progress
and inform the user, as the download may take time.
```javascript
const options = {
expectedInputLanguages: ['en'],
};
const available = if (Proofreader.availability("downloadable") === true);
```
> [!TIP]
> **Tip:** Use the [@types/dom-chromium-ai](https://www.npmjs.com/package/@types/dom-chromium-ai) npm package to get TypeScript typings for the Prompt API and other built-in AI APIs for easier access in your codebase.
To trigger the download and instantiate the proofreader, check for
[user activation](https://developer.chrome.com/docs/ai/get-started.md.txt#user-activation). Then, call the
asynchronous `Proofreader.create()` function.
```javascript
const session = await Proofreader.create({
monitor(m) {
m.addEventListener('downloadprogress', (e) => {
console.log(`Downloaded ${e.loaded * 100}%`);
});
},
...options,
});
```
### Create a Proofreader object
To create a Proofreader, use the `Proofreader.create()` function.
```javascript
const proofreader = await Proofreader.create({
expectedInputLanguages: ["en"],
monitor(m) {
m.addEventListener("downloadprogress", e => {
console.log(Downloaded ${e.loaded * 100}%);
});
}
};
```
The `create()` method includes the following options:
- `expectedInputLanguages`: An array of expected input languages.
The `includeCorrectionTypes` and `includeCorrectionExplanation` options from
the [explainer](https://github.com/webmachinelearning/proofreader-api?tab=readme-ov-file#detailed-design-discussion)
aren't supported.
### Start proofreading user text
Call `proofread()` to get corrections for an input text:
```javascript
const proofreadResult = await proofreader.proofread(
'I seen him yesterday at the store, and he bought two loafs of bread.',
);
```
Corrections are a type of `ProofreadResult`. Find the fully corrected input in
the `correctedInput` attribute and the list of corrections in the `corrections` array:
```javascript
let inputRenderIndex = 0;
console.log(proofreadResult.correction);
for (const correction of proofreadResult.corrections) {
// Render part of input that has no error.
if (correction.startIndex > inputRenderIndex) {
const unchangedInput = document.createElement('span');
unchangedInput.textContent = input.substring(inputRenderIndex, correction.startIndex);
editBox.append(unchangedInput);
}
// Render part of input that has an error and highlight as such.
const errorInput = document.createElement('span');
errorInput.textContent = input.substring(correction.startIndex, correction.endIndex);
errorInput.classList.add('error');
editBox.append(errorInput);
inputRenderIndex = correction.endIndex;
}
// Render the rest of the input that has no error.
if (inputRenderIndex !== input.length){
const unchangedInput = document.createElement('span');
unchangedInput.textContent = input.substring(inputRenderIndex, input.length);
editBox.append(unchangedInput);
}
```
## Permission Policy, iframes, and Web Workers
By default, the Proofreader API is only available to top-level windows and to
their same-origin iframes. Access to the API can be delegated to cross-origin
iframes using the Permission Policy allow="" attribute:
```html
```
The Proofreader API isn't available in Web Workers. This is due to the
complexity of establishing a responsible document for each worker, in order to
check the Permissions Policy status.
## Demo
Play around with the [Proofreader API playground](https://chrome.dev/web-ai-demos/proofreader-api-playground) and check out the
[source code](https://github.com/GoogleChromeLabs/web-ai-demos/tree/main/proofreader-api-playground).
## Engage and share feedback
The Proofreader API is under active discussion and subject to change in the future. If you try this API and have feedback, we'd love to hear it.
- [Read the explainer](https://github.com/explainers-by-googlers/proofreader-api), raise questions and participate in discussion.
- Review the implementation for Chrome on [Chrome Status](https://chromestatus.com/)
- Read the Mozilla Standards position and WebKit Standards position.
- [Join the early preview program](http://goo.gle/chrome-ai-dev-preview-join) for an early look at new APIs and access to our mailing list.
- If you have feedback on Chrome's implementation, file a [Chromium bug](https://new.crbug.com/).
Discover all of the [built-in AI APIs](https://developer.chrome.com/docs/ai/built-in-apis.md.txt) which use models, including large language models, in the browser.