Discussion: “Translator API not available — where can I get it?”

[Suraj Rana]

Hey everyone 👋

I need a Translator API for my project, but I can’t find it anywhere.
I checked this Chrome Developer Origin Trials page:
👉 https://developer.chrome.com/origintrials/#/trials/active
…but there’s no Translator API listed there.

Can anyone please tell me where I can get the Translator API or how to enable it?
If it’s not available through Chrome Origin Trials, is there any other official way or API service (like Google Cloud, Microsoft Translator, or DeepL) that I can use instead?

Any guidance or setup link would be really helpful 🙏

Thanks!

Suraj Rana

[Thomas Steiner]

Hi Suraj,

The Translator API shipped in Chrome 138. See this article for details on how to use it. 

Cheers,

Tom 

--
You received this message because you are subscribed to the Google Groups "Chrome Built-in AI Early Preview Program Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to chrome-ai-dev-previe...@chromium.org.
To view this discussion visit https://groups.google.com/a/chromium.org/d/msgid/chrome-ai-dev-preview-discuss/8c7790af-a3dd-45c9-ba28-fab0f2051efan%40chromium.org.

--

Thomas Steiner, PhD—Developer Relations Engineer (blog.tomayac.com, toot.cafe/@tomayac)

Google Spain, S.L.U.

Torre Picasso, Pl. Pablo Ruiz Picasso, 1, Tetuán, 28020 Madrid, Spain

CIF: B63272603

Inscrita en el Registro Mercantil de Madrid, sección 8, Hoja M­-435397 Tomo 24227 Folio 25

----- BEGIN PGP SIGNATURE -----

Version: GnuPG v2.4.8 (GNU/Linux)

iFy0uwAntT0bE3xtRa5AfeCheCkthAtTh3reSabiGbl0ck

0fjumBl3DCharaCTersAttH3b0ttom.xKcd.cOm/1181.

----- END PGP SIGNATURE -----

[Suraj Rana]

Hello Chrome AI Team,

  I am a participant in the "GOOGLE CHROME BUILT-IN AI CHALLENGE 2025" and am very excited to build with the new on-device AI APIs.

  My project is a writing assistant extension called "Writter Buddy" that aims to use Gemini Nano for tasks like proofreading and summarizing directly in the browser.

  I am running into an issue where the chrome.ai object is not available on my machine. Here are the details of my setup and the steps I've taken:

   * Machine: Apple iMac with M4 chip

   * Browser: The latest version of Chrome Canary

   * Flags Enabled:

       * #optimization-guide-on-device-model (set to Enabled BypassPerfRequirement)

       * #prompt-api-for-gemini-nano (set to Enabled)

  Despite this setup, when I check the DevTools console, chrome.ai returns undefined. This prevents the extension from initializing the AI session.

  I understand this is an early preview and access might be staged. Is it possible to have the chrome.ai feature enabled for my account so I can proceed with my hackathon project? Any guidance or help would be greatly appreciated.

  Thank you for your time and for this amazing technology. 

[Suraj Rana]

my account is - surajran...@gmail.com

[Connie Leung]

The apis  are enabled on the browser amd OS that you use. They are not tied to a Gmail account.

Maybe thomas knows whether or not built-in AI is supported on Imac

Connie 

To view this discussion visit https://groups.google.com/a/chromium.org/d/msgid/chrome-ai-dev-preview-discuss/faf4fbe7-d9f6-4ce2-bf3d-75d4d0fadcc4n%40chromium.org.

[Thomas Steiner]

Hi Suraj,

Please see the documentation of the various APIs that you can use for this challenge. It's all linked in the table at the beginning. All APIs are either shipped, so you can use them immediately, or are in origin trial, which means you need to sign up for them.

Cheers,

Tom

[Thomas Steiner]

Please see the hardware requirements in this expandable. Your machine definitely qualifies.

[Suraj Rana]

Hi Tom,

 Thank you for the quick reply and the clarification about Origin Trials.

It seems the on-device API is not yet enabled for my machine, as chrome.ai is undefined even in Canary with the correct flags.

I would like to build a true "Hybrid AI" project as mentioned in the hackathon rules. My proposed logic is:

   1. The extension will first check if chrome.ai is available.

   2. If it is, it will use the on-device Gemini Nano model.

   3. If it's not available, it will fall back to using the standard Google Gemini API (from 

    Google AI Studio) to perform the task.

  Would this be a valid "Hybrid AI" submission? This approach would allow the extension to

  work for all users while still prioritizing the on-device API when available.

  Thanks again,

  Suraj

[sharmaine erika delgado]

Ooh, I was facing the same dilemma earlier, and I had to sign up for the origin trial. Then I received the tokens.

To view this discussion visit https://groups.google.com/a/chromium.org/d/msgid/chrome-ai-dev-preview-discuss/CALgRrLn2GdOJVG%3Dg9v2_aTw9xO_8t__K2d3qj0g3tfUUW1qH%3DQ%40mail.gmail.com.

[Thomas Steiner]

Hi again,

Instead of `chrome.ai` you should use `window.LanguageModel`, `window.Summarizer`, etc. For a Chrome extension example, see this deeplink. Using the built-in APIs when they are available and then falling back to the cloud would be a valid hybrid approach, yes.

Cheers,

Tom

[José luis]

Solo dille a tu asistente que se sincronize dando instrucciones de traducción 

To view this discussion visit https://groups.google.com/a/chromium.org/d/msgid/chrome-ai-dev-preview-discuss/CALgRrL%3Di6Cm5gnRyz5aAnbZZpLxi6C4wJd3%2BKuJHBkB546hyZg%40mail.gmail.com.