Opera introduces the WebAI web standard for running AI models directly in the browser

The new WebAI standard

Opera, the browser developer, has introduced the WebAI web standard, which makes it possible to run AI models directly in the browser without turning to the cloud. The news was reported by TechCrunch. The standard is not tied to any specific model and is designed for a wide range of open-weight systems: Opera cites Google's open Gemma model as an example, and on Mac computers WebAI supports Apple's on-device foundation model.

How it works and where it is available

WebAI grew out of Opera's earlier projects to bring locally running models into its browsers. The new AI features are available for now as an early preview in the flagship Opera One browser; according to the company, the Opera GX browser for gamers and the agentic Opera Neon will receive them later. The feature is enabled only with the user's explicit consent. According to Opera, WebAI makes it possible to generate text, classify content, and find connections between data right in the browser.

Privacy and limitations

Opera emphasizes that open-weight models are usually free, their weights are publicly available, and data processing happens on the user's device — without sending personal information to the company's servers and without a subscription to cloud AI. “Developers can add AI features without running AI infrastructure or paying for each request,” the company's announcement says. The privacy context is relevant: as TechCrunch notes, the growing capabilities of AI assistants have already raised concerns about security and privacy, and privacy has long been among Opera's main selling points. At the same time, local execution involves a trade-off: it increases the load on the computer, and model performance will depend on the device's hardware.