Opera Launches WebAI Standard to Run AI Locally in the Browser

Most of the AI features we use daily have worked the same way: you type your request, it's sent to a remote server, and the answer comes back to your screen. This seemingly simple cycle carries a twofold cost: a financial cost borne by the developer or user in the form of subscriptions, and a privacy cost, because everything you type leaves your device. Opera is now proposing a different approach through a new web standard called WebAI, which allows AI models to run inside the browser itself, on the user's device, without passing through the cloud.

What is the WebAI standard?

WebAI is not an AI model, but rather an interface and technical standard that allows web pages and applications to access models running locally on the device. The standard was designed to be model-agnostic, meaning it isn't tied to any single model but supports a broad range of open-weight models. Opera pointed to Google's Gemma model as one example, and also noted that Mac users can take advantage of Apple's on-device foundation model.

This move is no sudden leap, but an extension of a path the company began earlier when it enabled local models to run within its browsers, then refined the experience into a standard that developers can adopt.

Where can you try it?

Opera has made the feature available as an early preview in its flagship browser, Opera One, with plans to bring it to its other releases later:

• Opera GX: the browser aimed at gamers and live streamers.
• Opera Neon: the browser that relies on AI agents to carry out tasks on the user's behalf.

The company stresses that enabling the feature is entirely optional: it won't run automatically without the user's explicit consent — an important point amid the growing debate over AI features being imposed on users without permission.

Why open-weight models?

The choice of open-weight models is no passing detail, as it offers several practical advantages:

• Usually free to use: they can be downloaded and run without per-request fees.
• Customizability: having the model's weights publicly available gives developers greater flexibility to adjust and adapt its behavior.
• Less reliance on a single provider: which reduces technical lock-in to any one company.

It is important, however, to be precise about the terminology: open-weight models are not open-source software in the full sense, as they typically come with licensing restrictions that define the permitted uses.

Privacy at the heart of the idea

Opera sees privacy as the most significant gain. When processing takes place on the user's device, the data the model handles stays local and is never sent to external servers. According to the company's announcement, developers can add AI features without running costly infrastructure or paying per query, while users get the same functionality without subscriptions and with greater control over what happens on their own devices.

This point matters because AI assistants now receive a great deal of personal information, and many users share sensitive details without a clear understanding of how that data is stored or later used. Local processing doesn't eliminate every risk, but it returns part of the decision to the user's hands — an approach consistent with the position Opera has sought to establish for itself over the years as a browser focused on digital privacy.

Available capabilities and trade-offs

The company notes that the standard enables a variety of tasks to be carried out directly in the browser, including text generation, content classification, and analyzing relationships between different pieces of information.

On the other hand, there is a clear price to this approach: running models locally consumes device resources in terms of processing power and memory, which may affect overall performance. Opera openly acknowledges that the quality and speed of results will depend on each user's hardware specifications, meaning the experience will vary widely between a modern device and one with modest capabilities.

Reading the broader trend

What Opera is offering reflects a wider shift in the AI industry toward distributing computation between the cloud and edge devices, rather than concentrating it entirely in data centers. As smaller models become more efficient and personal processors grow more capable, running some tasks locally becomes a realistic option rather than merely a technical experiment.

The standard's success will ultimately depend on how widely developers adopt it and whether other browsers support it, since open standards only prove their worth when they move beyond the boundaries of the company that launched them.

✦ بقلم فريق دروب أيديا