Opera has started rolling out an opt-in preview that lets websites and extensions use AI models running directly on users’ computers.
Opera began rolling out local AI models in its browsers on October 27, introducing an early preview of WebAI in Opera One. The feature lets websites and browser extensions use AI hosted by the browser, rather than sending requests to a remote AI service. Opera says Opera GX and Opera Neon are expected to follow.
WebAI is a proposed web standard, not an Opera-only model. Opera’s implementation is model-agnostic and supports open models; at launch, users can access Google’s Gemma and, on Macs, Apple’s on-device foundation model. The browser can use these models to generate text, classify content and identify relationships between information. Opera says developers can build such features without operating AI infrastructure or paying for each request.
The preview is opt-in: models are not downloaded until users enable the feature, and site-specific permissions let them decide which websites can access local AI. Availability is initially limited to 64-bit Windows computers with Intel or AMD processors and Macs with Apple silicon. Processing takes place on the user’s device, so performance depends on its hardware.
The launch brings local-model access from an experimental developer build Opera introduced in 2024 into its consumer browser. It also gives extension developers a way to experiment with AI features that can work without sending the handled information to an AI service. Opera’s examples include extensions that sort tabs or flag clickbait, though these are demonstrations rather than announced built-in browser functions. The company says it plans to add more models optimized for different devices and additional development tools. The broader significance will depend on which models Opera supports, how well they run across different computers, and whether developers build useful extensions around the permission-based approach.