Opera Launches WebAI to Run AI Models Directly in Browsers

Browser manufacturer Opera has unveiled a novel web standard that enables users to execute AI models directly within the browser interface, eliminating the need to rely on cloud infrastructure.

This new standard, WebAI, is engineered to function with a diverse array of AI models. Opera cites Google's Gemma—an open-weight model introduced two years ago—as a primary example of compatibility. Additionally, the company reports that WebAI supports Apple's on-device foundation models on Mac computers.

The company emphasizes that WebAI is not limited to a specific model and is architected to accommodate a broad spectrum of open-weight AI systems. This standard evolved from previous Opera initiatives aimed at integrating locally executed models into its browsing products.

Opera is currently rolling out these AI capabilities as an early preview within its flagship Opera One browser. Furthermore, the company indicates that Opera GX, a browser tailored for gamers and streamers, and Opera Neon, which utilizes AI agents for task execution, will soon receive similar updates. Users are required to actively opt in to enable these features, the company states.

Open-weight models provide several advantages to users. Primarily, they are generally free to download and utilize. Furthermore, the underlying model weights are publicly accessible, granting developers greater freedom to customize functionality. However, unlike fully open-source software, these models still carry certain restrictions.

A significant additional benefit is enhanced privacy, as noted by the company. Users can access AI functionalities without transmitting their personal data to third-party servers or purchasing expensive cloud-based AI subscriptions.

“Developers can incorporate AI features without maintaining AI infrastructure or paying for each individual request,” the company stated in its announcement. “Users do not require AI subscriptions; the information processed by the local AI remains on their own device, granting them continued control over the AI on their computer.”

Opera also revealed that WebAI enables users to generate text, classify content, and comprehend relationships between information fragments directly through the browser.

Privacy has emerged as a growing concern as AI assistants become increasingly powerful and gain access to personal data. Users frequently relinquish vast amounts of personal details to AI chatbots, often without fully grasping how that data is stored or utilized.

Locally executed models provide an alternative solution since processing occurs on the user's own device rather than on a company's cloud server. While this does not eradicate every privacy risk, it offers users greater control over their data. It is fitting that Opera has long emphasized digital privacy as a core selling point.

Nevertheless, there are trade-offs involved. Executing AI models locally places greater strain on a user's computer, potentially impacting performance. Opera notes that the performance of these models will vary depending on the device's specific hardware specifications.