Your browser is about to become an AI computer. Opera just shipped WebAI, a new web standard that runs AI models directly inside the browser: no subscriptions, no cloud, no data leaving your machine. Google and Microsoft are backing the standard, the early preview is live right now, and it might be the easiest local AI browser setup anyone has shipped yet.
What WebAI actually is
WebAI is a proposed web standard that bakes local AI inference into the browser itself. In plain English: the AI model runs on your computer, inside Opera, and websites or extensions can use it through the browser.
Opera built the whole pipeline into the browser, so the model download, the memory management, and the interface between the model and web pages all happen behind the scenes. There’s no separate app, no server config, and no API key to beg for.
Opera actually pioneered this space. Back in 2024 it was the first major browser to ship built-in access to local AI models, though only in an experimental developer build. WebAI is the grown-up version of that experiment, promoted to a real standard that other companies are rallying behind. Opera’s announcement names Google and Microsoft as companies backing the proposed standard, which tells you where the industry is heading: browsers want to be AI platforms.
The standard isn’t tied to one model, either. Opera highlights Google’s open-weight Gemma family as the flagship example, and on Macs, WebAI also taps Apple’s on-device foundation model. Because it supports any compatible open-weight model, you’re not married to one company’s AI. You pick a model, the browser runs it, done.
Why running local AI in your browser matters
Cloud AI has three costs, and you pay all of them.
Money first. ChatGPT Plus, Gemini Advanced, Claude Pro, those subscriptions run $20 or more per month each. A local model in your browser costs nothing after the download. Zero per month. Your GPU does the work instead of a data center you rent by the token.
Privacy second. Every chat with a cloud AI sends your words to somebody’s servers. Most people are fine with that for recipes. It hits different when you’re pasting contracts, medical questions, or work documents into the box. With a local AI browser setup, the information never leaves your device. Opera’s pitch says it directly: the data the local AI handles stays on your own machine.
Control third. Cloud providers rate-limit you, change their models without asking, and deprecate features you liked. Local models do none of that. The model you downloaded today behaves the same way next year.
If you’ve followed this site for a while, you know we like this category. We’ve covered how to turn an old PC into a local AI server and the best offline tasks for a local LLM on your phone. WebAI is the same idea applied to the place you already spend your day: the browser. No spare hardware required.
What you can do with it today
The first release is an early preview in Opera One, with Opera GX (the gamer-focused browser) and Opera Neon (the agentic one) getting it soon. You have to opt in, so nothing changes until you flip the switch yourself.
What does the preview actually do? Per Opera’s announcement, WebAI lets the browser generate text, classify content, and understand how pieces of information relate to each other. Translated into real tasks:
- Summarize a long article or a wall of comments without copy-pasting into a chatbot
- Draft and rewrite text right on the page you’re typing on
- Classify and sort things, like tagging a pile of listings or messages
- Answer questions about the page you’re reading, with no round trip to a server
Early preview means rough edges. Don’t expect the polish of a flagship cloud model. Expect a genuinely useful assistant that works on a plane, works on the family’s aging desktop, and never asks for a credit card.
Here’s the honest comparison of where each option stands:
| Cloud AI (chatbots) | WebAI in Opera | |
|---|---|---|
| Monthly cost | $20+ per subscription | Free |
| Where data goes | Company servers | Stays on your device |
| Works offline | No | Yes, after model download |
| Model quality | Frontier-level | Good for small tasks, not frontier |
| Setup effort | Sign up | Download model in browser |
| Rate limits | Yes | Limited by your hardware only |
The catch: your hardware does the work
Nothing is free, and local AI’s price is paid in hardware. Running a model in your browser uses your CPU, your RAM, and sometimes your GPU. On a decent modern laptop, that’s fine. On a ten-year-old budget machine, expect slowness and fan noise.
Opera is upfront that model performance depends on the device. The bigger the model, the smarter the output and the heavier the load. Small models handle summaries and drafts well. Complex reasoning is where they lag behind the cloud flagships, so keep expectations calibrated: a local model is a capable intern, not a senior consultant.
There’s also a storage angle. Models take gigabytes, so the first download isn’t instant, and your SSD gives up some room. Worth it if you use AI daily. Skip it if you generate text twice a month.
One more honest note: local means private, not anonymous. If a website sends your data somewhere else through its own forms, WebAI doesn’t stop that. It keeps the AI processing on your machine. Regular web privacy habits still apply, and websites blocking AI agents is a separate fight happening right now.
How to try Opera’s local AI browser
Curious? Here’s the quick path.
- Download Opera One from opera.com (it’s free, and it runs on Windows, Mac, and Linux).
- Open Settings and look for the AI section, where the WebAI early preview lives.
- Opt in, then pick a model to download, Gemma is the default suggestion.
- Wait for the download to finish, then test it on whatever page you’re reading: ask for a summary and watch it answer without the internet round trip.
On a Mac, the browser can also use Apple’s on-device model, which is tuned nicely for Apple silicon. Opera GX and Opera Neon users get the same features in an upcoming update, so gamers and agent-power-users aren’t left out for long.
What this means for Chrome users
The standard is backed by Google and Microsoft, which is the real headline hiding in this announcement. When the companies behind Chrome and Edge co-sign a local AI web standard, in-browser AI stops being an Opera experiment and becomes a platform shift. Chrome already ships Gemini features, and Edge bolts Copilot into every corner. The standard gives every browser a common, private way to do the same thing locally.
So even if you never install Opera, the local AI browser features you’ll eventually get in Chrome or Edge probably started here. And for people who want AI help without another subscription, that’s the best news of the week.
Takeaway
Opera’s WebAI turns Opera One into a free local AI browser, and the early preview is live today. If you’ve been curious about local AI but scared off by terminal commands and GPU shopping, this is the easiest on-ramp yet: install, opt in, download a model, and let your browser do the thinking. Try it on a page you’re reading right now and see how much you can get done without the cloud.