WebAI

WebAI is Opera's name for in-browser local AI: sites and extensions can call models that run on the user's computer—no cloud round-trip, no separate AI app, and no vendor API key required. Opera's blog announced an early preview in Opera One / GX / Neon on 2026-10-27; 2026-10-28 brought a Product Hunt listing and TechCrunch coverage. Blog: Opera News — WebAI · Docs: next.opera.com/web-ai · Settings → Opera AI → Enable Web AI API.

Key points

1. Pitch: local inference built into the browser so page content need not leave the device. 2. APIs: Prompt, Decision, and Semantic Embeddings for generation, classification, and relatedness. 3. Models: open-weight models via Opera's engine plus compatible on-device models (including Google Gemma; Apple's on-device model on Apple silicon). 4. Control: opt-in; nothing downloads until enabled; can be turned off anytime. 5. Platforms: early preview on 64-bit Windows (Intel/AMD) and Apple-silicon Macs — not Intel Macs, Linux, or Windows on Arm. Speed depends on hardware. 6. ** demos**: six experimental extensions (Project Lens, Got It?, Semantic Art, etc.).

Why this radar term

Concrete proper noun WebAI (not a generic "browser AI" phrase) with an official Opera post, press, and same-day PH discussion; not previously in this radar DB.

Get started

1. Install/update Opera One. 2. Settings → Opera AI → enable Web AI API (model download may take a few minutes). 3. Follow next.opera.com/web-ai or try the sample extensions.

Tips

• Still an early preview — APIs and model catalog may change. - Local inference taxes the device; low-end hardware will feel slow. - "Web standard" status is proposed; cross-browser adoption is TBD.