Opera delivers local AI models directly in browsers

On October 27, 2026, Opera launched the WebAI web standard in the early preview of Opera One, providing sites and extensions with three APIs for working with local AI models: Prompt, Decision, and Semantic Embeddings.

Compiled by Sergey KostenchukPublished 2026-10-28Updated 2026-10-28

2026-10-28 Coding Google

🛠 Opera: Local AI models in browsers

On October 27, Opera launched WebAI in the early preview of Opera One — a web standard supported by Google and Microsoft. Three APIs for sites and extensions: Prompt (text generation), Decision (classification), and Semantic Embeddings (finding connections). At launch — the open Gemma model and an Apple model on Mac. Nothing is downloaded until the feature is enabled in settings, and access is granted separately to each site.

🌍 Part of the AI workload is moving from cloud APIs to the device for the first time — extension developers have access to local models without per-request fees.

👤 Enable the Web AI API in Opera One settings and try the demo extensions Project Lens, Got It?, and Semantic Art: text generation and classification — locally, without subscriptions, data does not leave the computer.

Source 1: https://press.opera.com/2026/10/27/opera-doubles-down-on-local-ai-shipping-models-directly-in-its-browsers/

Source 2: https://next.opera.com/blog/web-ai-local-ai/

Sources