Opera is previewing WebAI, a new web standard designed to let websites and browser features run AI models directly on a user’s computer rather than sending requests to the cloud, TechCrunch reports.
The standard is available as an early preview in Opera One. Opera says it can work with different AI systems, including Google’s Gemma open-weight model and Apple’s on-device foundation model on Macs. The company says WebAI is not tied to a single model family and is intended to support a broad range of open-weight systems.
Local processing and privacy
Opera says WebAI can support tasks such as text generation, content classification and identifying relationships between pieces of information. Because processing takes place on the local device, users can access those features without sending the handled information to a company’s servers or paying for a cloud AI subscription.
Local execution can offer more control over personal data, but it does not remove every privacy risk. Open-weight models are generally available for download and customization, although they can still carry usage restrictions and are not necessarily fully open-source.
Hardware remains a limitation
Running models locally also places greater demands on a computer’s processor, memory and other hardware. Opera says performance will vary depending on the device.
The company plans to bring the feature to Opera GX, its browser for gamers and streamers, and Opera Neon, its AI-focused browser. Users must opt in to the preview. For developers, Opera says the approach could make it possible to add AI capabilities without operating model infrastructure or paying for each cloud request.