Browser maker Opera has launched a new web standard called WebAI, allowing users to run AI models directly in the browser without relying on the cloud.
Opera emphasizes that WebAI is not locked to a single model, but built for a variety of open-weight systems. They cite Google's Gemma as an example, an open-weight model that debuted two years ago; on Mac, WebAI can also access Apple's own on-device foundation models.
Currently, WebAI has appeared in early preview form in Opera One, the flagship product. Opera GX, aimed at gamers and streamers, and Opera Neon, which can get work done through agents, will soon follow. However, users must actively enable the feature for it to take effect.
Dual selling points: privacy and cost
The benefits of open-weight models are very tangible: they can usually be downloaded for free and used freely, and their public weights give developers greater room to customize to their needs, though not quite fully open source, usage and modification remain restricted.
Opera puts privacy front and center: when using AI, there is no need to send personal information to corporate servers, nor to pay for cloud service subscriptions. The announcement states it plainly—developers can add AI to their products without building their own AI infrastructure or paying per use; users don't need to buy subscriptions, information processed locally stays on their own devices, and the AI on their computer remains under their own control. With WebAI, users can generate text, classify content, and understand relationships between pieces of information directly in the browser.
Local models are not risk-free, but they offer more control. The trade-off is that they directly consume local computing power and may drag down performance, with the actual experience depending on hardware configuration. As AI assistants can read more and more personal information, privacy concerns are rising, and keeping all processing on the local machine rather than handing it to corporate clouds aligns precisely with Opera's long-standing focus on digital privacy.