Gemma 3n Model

[Scott Fortmann-Roe]

Great to hear about the updated Gemma 3n based model.

Two brief questions:

How do we know if we have the updated model? Is it "v3Nano" in the <chrome://on-device-internals/>?

Will this have an impact on VRAM and disk space requirements?

[Thomas Steiner]

I've asked Kenji to respond to this question on the announcement list where you have asked the same. Please wait for him to get online again. Thanks for the patience. 

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 -----

--
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/CAHqtauFtcse63jZE0Saa9_XyPmTn4%2B1FNLf_jcJMd1mBcSF%3D7g%40mail.gmail.com.

[Scott Fortmann-Roe]

Oh, sorry for asking twice. I got an auto-bounce response on my reply to the announcement email and thought it hadn't gone through.

[Jacob Kinnaird]

I don't even know what these are

Get Outlook for Android

---

From: 'Scott Fortmann-Roe' via Chrome Built-in AI Early Preview Program Discussions <chrome-ai-dev-...@chromium.org>
Sent: Friday, July 25, 2025 3:19:21 AM
To: Thomas Steiner <to...@google.com>
Cc: Chrome Built-in AI Early Preview Program Discussions <chrome-ai-dev-...@chromium.org>
Subject: Re: [chrome-ai-dev-preview-discuss] Gemma 3n Model

 

To view this discussion visit https://groups.google.com/a/chromium.org/d/msgid/chrome-ai-dev-preview-discuss/CAHqtauGhvYA1kqf53XwOg5gf_tYEEb4HbdK7tnY5xDUuy7HPCg%40mail.gmail.com.

[Thomas Steiner]

No worries, it is confusing. The owners of the announcement mailing list at Google receive the message, even if it bounces for you. If we then reply, we're allowed to post to everyone as we're owners, which has caused some confusion for you all except for the original sender, as you didn't have the context of the seemingly bounced email. We now work with an automated labeling system in Gmail so hopefully this human error of us replying by accident to bounced emails doesn't happen again. Apologies for the inconvenience. 

We do reply-all on this discussion mailing list. Be sure to make use of the "mute" feature in Gmail if you're no longer interested in a thread. 

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 -----

[Jacob Kinnaird]

I apologize for the confusion.  I've had a confusing week. 

Did I miss something?

Get Outlook for Android

---

From: 'Thomas Steiner' via Chrome Built-in AI Early Preview Program Discussions <chrome-ai-dev-...@chromium.org>
Sent: Friday, July 25, 2025 3:39:54 AM
To: Scott Fortmann-Roe <sc...@blaze.today>
Cc: Thomas Steiner <to...@google.com>; Chrome Built-in AI Early Preview Program Discussions <chrome-ai-dev-...@chromium.org>

Subject: Re: [chrome-ai-dev-preview-discuss] Gemma 3n Model

To view this discussion visit https://groups.google.com/a/chromium.org/d/msgid/chrome-ai-dev-preview-discuss/CALgRrLkMPGa08jPmzswr0-ca72rxmj4n4J34ycNYeGXR1898mA%40mail.gmail.com.

[Die4Ever2005]

I think it might be Version: 2025.06.30.1229 in chrome://on-device-internals/ Model Status. I'm seeing inputQuota of 9216 instead of the old 6144.