Opera Ships Local AI Models in Its Browsers
Opera on Oct. 27 began shipping local AI models in Opera One through WebAI, a set of browser APIs that let websites and extensions run AI features on the user's own computer. Opera GX and Neon follow.
Opera on Oct. 27 began shipping local AI models inside its browsers, letting websites and extensions run language-model features on a user's own computer through a set of browser APIs it calls WebAI. The early preview is live in Opera One now, with Opera GX and Opera Neon to follow, the Oslo-based company said in a press release.

Key takeaways
- WebAI exposes browser-hosted AI to web pages and extensions, so developers need no AI servers, API keys or per-request fees, according to Opera.
- At launch, users can run Google's Gemma models, plus Apple's on-device foundation model on Macs.
- The feature is opt-in: nothing downloads until a user turns it on, and site-by-site permissions control access.
What Opera WebAI does
Opera describes WebAI as a proposed web standard backed by companies including Google and Microsoft. Its own implementation is model-agnostic and built around open models, so users are not locked into one AI provider.
Opera's developer page lists eight interfaces: the Prompt, Summarizer, Writer, Rewriter, Translator, Proofreader, Semantic Embedder and Decisions APIs. Together they cover text generation, content classification and finding relationships between pieces of text.
The company gave a simple example of what that enables: an extension that sorts tabs, flags clickbait or spots relevant job postings using a model that runs on the laptop rather than in a data center.
How to turn on local AI in Opera
Users enable the Web AI API under Settings, in the Opera AI section, which downloads a compatible local model. Opera says the download may take a few minutes and that no separate app, local server or API key is required.
Hardware support is limited for now. WebAI runs on 64-bit Windows PCs with Intel or AMD processors and on Macs with Apple silicon. Intel-based Macs, Linux and Windows on Arm are not supported, and speed depends on the user's machine.
To seed development, Opera published six experimental extensions it calls examples rather than product features. They include Project Lens, which colors an article by how relevant each paragraph is to a user's project, and Got It?, which writes a quiz about the page just read.
From developer experiment to default feature
The launch builds on an April 3, 2024, experiment in the Opera One developer stream, which offered about 150 local model variants from roughly 50 model families. Each variant typically needed 2 to 10 gigabytes of storage, Opera said at the time.
"We first shipped local models to developers in 2024. Now we're bringing them to everyone," said Krystian Kolondra, Opera's executive vice president for browsers and AI. "Your laptop can already do it, without sending your data anywhere."
Opera is not alone. Google's Chrome for Developers documentation lists Translator, Language Detector and Summarizer APIs as available in stable Chrome since version 138, running on a built-in model. Opera's twist is letting users pick among open models instead of a single bundled one.
Opera also launched HTML-in-Canvas on the same day, a feature that places working web pages inside 2D and 3D scenes, aimed largely at games and Opera GX.
What's next
Opera says it plans to add more open models optimized for different devices and more tools for building on local AI, alongside the GX and Neon rollout. Whether WebAI becomes a cross-browser standard will depend on how far the proposal advances with other browser makers.
FAQ
Does Opera WebAI send my data to the cloud?
No. Opera says the models run on the user's device, and data stays there. Websites need permission to use local AI.
Which computers can run it?
64-bit Windows PCs with Intel or AMD chips and Apple silicon Macs. Intel Macs, Linux and Windows on Arm are not supported yet.
Do developers pay to use it?
Opera says local AI lets developers add AI features without running infrastructure or paying per request.
References
- Opera doubles down on local AI, shipping models directly in its browsers — Opera (Oct. 27, 2026)
- WebAI: Local AI for the web, now in Opera — Opera blog (Oct. 27, 2026)
- Opera becomes the first major browser with built-in access to local AI models — Opera (April 3, 2024)
- Built-in AI APIs — Chrome for Developers
Editor responsible for AI Newsway. Writes about AI, developer tools, cloud and productivity technology.
Comments 0
Comments are currently disabled.