Browser developer Opera has unveiled the new WebAI web standard, aiming to let users run AI models directly in their browsers without depending on cloud-based services. The standard supports various AI models; Opera used Google’s open-source Gemma model as an example and noted that WebAI also works with Apple’s on-device foundation models. This technology does not rely on any particular model and is primarily intended for diverse open-weight AI systems, continuing Opera’s previous work exploring local execution of large language models in browsers.
At present, WebAI is available as a preview in Opera One, Opera’s flagship browser, and users need to actively opt-in to enable it. Opera GX, aimed at gamers and streamers, as well as Opera Neon, which supports AI-powered task automation, will also receive related features in the future.
WebAI can generate text, classify content, and analyze relationships between pieces of information—all within the browser. According to Opera, local inference reduces reliance on cloud-based AI infrastructure and paid subscriptions while keeping user data on personal devices, thereby enhancing privacy controls. However, open-weight models are not equivalent to fully open-source solutions, and running models locally cannot completely eliminate privacy risks. Moreover, actual performance still depends on the device’s hardware configuration.
As AI assistants increasingly process personal information, edge AI is emerging as an important alternative to cloud-based services. WebAI seeks to further integrate model-running capabilities into browsers, but balancing privacy protection, device performance, and model capabilities remains a key challenge for the widespread adoption of local AI applications.