Opera built an inference engine directly into its browser to let websites and browser extensions run AI models locally on users' computers. The feature, called WebAI, launched as an opt-in early preview in Opera One.
WebAI provides three APIs for text generation, classification, and finding related content. Opera released six example extensions, including one that quizzes users on articles and another that highlights relevant paragraphs for projects.
The tool supports local models including Gemma and Apple's on-device foundation model on Macs. Opera did not describe how sites negotiate which local model they access or what permissions users grant to individual websites using WebAI.