Opera WebAI Lets Users Run Local AI Models Directly in Browser Without Cloud Dependency

Opera Launches WebAI to Enable On-Device AI Processing

Opera has introduced Opera WebAI, a proposed new web standard that allows users to run AI models locally in their browser without sending data to cloud services1

1

. The browser AI feature is now available as an early preview in Opera One browser, with Opera GX and Opera Neon scheduled to receive the same capabilities soon1

1

4

4

. The implementation is model-agnostic and supports open-weight AI models including Google's Gemma and Apple's on-device foundation model on Apple silicon Macs2

2

.

Privacy-Focused Alternative to Cloud AI Services

The WebAI standard, backed by companies including Google and Microsoft, represents a shift toward on-device AI processing that prioritizes data privacy and user control2

2

. Users can access AI features without surrendering personal information to company servers or paying for cloud-based AI subscriptions1

1

. Krystian Kolondra, EVP Browsers and AI at Opera, stated that much of what people want from AI doesn't require a data center and can already run on their laptops without sending data anywhere4

4

. The opt-in feature ensures nothing is downloaded until users enable it, and each website requires permission before accessing local AI3

3

.

Developer Access to Local AI Without Infrastructure Costs

WebAI opens local AI capabilities to third-party developers instead of reserving them exclusively for Opera's own features3

3

. Developers can add AI features without running AI infrastructure or paying for every request, while users don't need AI subscriptions1

1

. The platform provides eight built-in browser APIs including Prompt API for real-time chat, Summarizer API, Writer API, Rewriter API, Translator, Proofreader, Semantic Embeddings API for converting text into mathematical vectors, and Decisions API for multiple-choice sorting5

5

. Opera released six experimental extensions demonstrating practical applications such as Project Lens for context-based article highlighting, Got It? for reading comprehension quizzes, and Questionable Intelligence for chart-based data analysis4

4

.

Hardware Demands and Performance Considerations

Running AI models locally places increased hardware demands on user devices, which can affect performance1

1

. Opera notes that model performance depends on device hardware, though the company hasn't disclosed minimum performance requirements beyond supported platforms3

3

. The early preview is currently available on 64-bit Windows PCs with Intel or AMD processors and Macs with Apple silicon, excluding Intel-based Macs, as well as Linux and Windows on Arm4

4

. This trade-off between cloud services and local processing means users gain privacy and cost savings but must rely on their own computing power3

3

.

HTML-in-Canvas Feature Adds Interactive 3D Web Experiences

Alongside WebAI, Opera launched HTML-in-Canvas, which lets developers embed fully functional web pages inside 2D and 3D virtual scenes2

2

. Interactive elements such as menus, buttons, and forms remain usable even as they move within a scene, built using familiar HTML and CSS2

2

. Maciej Kocemba, Product Director at Opera GX, described taking real web pages and throwing them into 3D worlds where everything still works2

2

. Opera created two demonstrations: Webpage Bulldozer, which turns web page elements into physical objects users can manipulate, and Gamer Room, a virtual 3D room featuring screens displaying live web pages5

5

. Both demos work in Opera browsers without additional setup2

2

.