The Pulse
Opera Ships WebAI Preview for On-Device Browser Models
Opera has begun rolling out WebAI, a browser feature that lets websites and extensions use AI models running on a user’s device. Opera’s newsroom release and blog describe the initial browser availability differently, while both present Web
AI.info Team · Written with AI · fact-checked ·
Opera began shipping local AI in its browsers on October 27, introducing WebAI as a way for websites and extensions to use models running on a user’s device. The company calls WebAI a proposed web standard, not a finalized specification. Opera’s newsroom release says the first early-preview capabilities are available in Opera One, with Opera GX and Opera Neon to follow; a separate Opera blog describes the preview as already available across all three browsers.
WebAI puts the browser between sites and models
The proposal gives websites and browser extensions a route to browser-hosted AI for tasks such as generating text, classifying content and finding relationships between pieces of information. Opera says its implementation can work with different models rather than tying developers to one provider. At launch, the company lists Google’s Gemma and, on Macs, Apple’s on-device foundation model.
Opera presents the approach as a way to build browser features without operating AI infrastructure or paying a cloud provider for each request. Examples in its announcement include extensions that sort tabs, flag clickbait or identify relevant job postings. Those are potential uses, not built-in Opera features; the company says developers can create extensions to perform them.
Users must enable local models
WebAI is opt-in: Opera says it downloads nothing until a user enables the feature, and site-specific permissions let users choose which websites can access local AI. The company’s developer instructions direct users to download or update Opera One, then go to Settings and Opera AI to turn on the Web AI API. Enabling it downloads a compatible model, a step that Opera says may take a few minutes.
The preview supports 64-bit Windows PCs with Intel or AMD processors and Macs with Apple silicon. Opera’s blog excludes Intel-based Macs, Linux and Windows on Arm. Because inference runs on the user’s machine, response speed depends on that device’s hardware, the company says.
Opera’s pitch: fewer cloud calls, more user control
“A lot of what people want from AI doesn’t need a data center. Your laptop can already do it, without sending your data anywhere.”
Krystian Kolondra, EVP Browsers and AI at Opera
Local processing can keep the content handled by a model on the device, avoiding a trip to an external AI service for that inference. It also shifts the computing work to the user’s own hardware, so model performance will vary across machines. Opera says the preview is the first step in a larger effort to add models tuned for different devices and more tools for building browser features.
A proposal, not a cross-browser standard yet
Opera says WebAI is backed by companies including Google and Microsoft, but the announcement does not describe a finalized standard or say that other browsers have implemented it. For now, the feature is Opera’s early preview, with supported models, platforms and permissions defined by the company’s implementation. Whether WebAI becomes a common interface across browsers depends on work beyond this initial release.
The immediate offer is narrower and concrete: developers can experiment with local text generation, classification and semantic relationships through Opera’s browser, while users decide whether to download a model and which sites can call it. Opera plans to add more open models optimized for different devices.