Browser maker Opera has introduced WebAI, a proposed web standard for running AI models directly in the browser without relying on the cloud. The feature is available as an early preview in Opera One, and users must opt in. Opera says WebAI is not tied to any particular model and is built to support a wide range of open-weight AI systems; Google’s Gemma is cited as an example, and on Macs it also supports Apple’s on-device foundation model. The standard grew out of Opera’s previous work on locally run models, and Opera GX and Opera Neon are expected to get the same support soon.

The motivation is to offer an alternative to cloud-dependent AI. Open-weight models are generally free to download and use, and their publicly available weights give developers room to customize them, though they remain more restricted than fully open-source software. Opera also frames local inference as a privacy advantage: users can use AI features without sending personal information to company servers, paying for cloud subscriptions, or paying per request. The company says developers can add AI features without running AI infrastructure or paying for every request, while users stay in control of the AI on their computer. With the same local pipeline, the browser can generate text, classify content, and understand how pieces of information relate.

Local execution does not come without costs. Running models on-device adds load to the user’s computer, and Opera acknowledges that model performance will depend on the device’s hardware. The announcement sits in a broader context of growing privacy concerns around AI assistants, which often collect personal data without clear visibility into storage and use. Locally run models do not eliminate every privacy risk, but they shift more control to the user — a direction consistent with Opera’s long-standing emphasis on privacy.