Skip to content

Opera WebAI: Your Browser Can Now Run AI Locally, With No Cloud and No Subscription

On October 27, 2026 Opera announced WebAI, a proposed web standard that lets websites use AI models running on your own PC instead of the cloud. Here is how it works, which models it uses, and how to turn it on.

By Zeeshan Khalid · Updated 30 October 2026

Most AI features on websites send what you type to someone else's server. Opera wants to change that.

On October 27, 2026, Opera announced WebAI, a proposed web standard that lets websites and browser extensions use AI models running on your own computer. Your text stays on your device and no one charges you per request.

What WebAI actually does

WebAI gives web pages three new browser APIs for:

  1. Generating text (summaries, rewrites, answers),
  2. Classifying content (for example sorting or tagging),
  3. Analysing how documents relate to each other.

The browser runs the model on your own graphics card and memory. There are no per-request cloud fees for the website and no subscription for you.

Opera also released six experimental extensions to show what it can do, including one that highlights key paragraphs on a page and one that turns an article into a quiz.

Which models it uses

WebAI runs open-weight models, including:

  • Google's Gemma, and
  • Apple's on-device foundation models on supported Macs.

Who can use it now

It is an early preview in Opera One on:

  • 64-bit Windows PCs, and
  • Apple Silicon Macs.

Opera GX and Opera Neon are coming "soon," with no date yet. You don't need to install anything else or run a local server.

Privacy: off by default

  • WebAI is turned off until you enable it in browser settings.
  • No models are downloaded until you switch it on.
  • Each website or extension must ask permission before it can use your GPU or memory.

As Opera's spokesperson put it: "WebAI stays completely turned off until users decide to enable it in their browser settings."

The catch

  • Speed depends on your hardware. A new laptop with a decent GPU will be fast; an old one will be slow.
  • It is a proposal, not a standard yet. Opera says Google and Microsoft support the idea, but it hasn't been formally approved. Until other browsers adopt it, websites can only rely on it for Opera users.
  • Small local models are not frontier models. They are great for summarising and sorting, but not for the hardest reasoning tasks.

Why it matters

Opera's argument is simple. In the words of executive Krystian Kolondra, "routine web tasks do not need massive data centers." If other browsers adopt WebAI, AI features on websites could become faster, cheaper and private by default.

Want local AI today, in any browser?

You don't have to wait for WebAI:

Try these next