“Developers can add AI features without running AI infrastructure or paying for every request,” as browser maker Opera stated in its announcement.
Browser maker Opera has introduced a new web standard called WebAI that lets users run AI models directly in its browser, without relying on the cloud. The transition to client-side execution shifts the financial burden of running large language models off corporate balance sheets entirely. By moving inference to the end user's device, software creators avoid the recurring operational expenditures tied to API calls and cloud server maintenance. For companies building applications, this architectural shift eliminates the core variable cost associated with artificial intelligence features. The elimination of server-side compute outlays changes unit economics for applications that previously faced heavy cloud infrastructure bills for every text generation or classification task. By offloading computation to the client, companies trade away predictable cloud performance for completely zero marginal infrastructure costs.
Shifting Infrastructure Outlays to Client Hardware
The new standard, WebAI, is designed to work with a variety of AI models. Opera highlights Google’s Gemma — an open-weight model introduced two years ago — as one primary example. On Macs, WebAI also supports Apple’s on-device foundation model, as the company noted. Opera points out that WebAI is not tied to any single proprietary model and is built to support a wide range of open-weight systems. WebAI grew out of previous work by Opera to bring locally run models directly into its desktop browsers.
Open-weight models offer specific operational benefits to developers. They are typically free to download and deploy, while their underlying weights remain publicly accessible, granting development teams tighter control over customization. Unlike fully open-source software, however, these models still carry distinct licensing restrictions.
Privacy serves as another core vector for the rollout, as the company stated. Users can tap into AI features without transmitting personal data to external corporate servers or funding a cloud-based AI subscription. Opera is making the new capabilities available as an early preview in its flagship Opera One browser. The company plans to expand the standard to Opera GX, its browser targeted at gamers and streamers, and Opera Neon, which utilizes autonomous AI agents for task execution. Users must manually opt into the feature, according to the company.
“Users don’t need AI subscriptions, the information the local AI handles stays on their own device, and they stay in control of the AI on their computer,” as the company commented.
According to Opera, WebAI also allows clients to handle text generation, content classification, and semantic analysis directly within the browser environment. Digital privacy has emerged as a central market differentiator as AI assistants expand their access to sensitive user data. Traditional cloud-based chatbots require users to surrender vast amounts of information without transparent storage or processing guarantees. Locally executed models mitigate this exposure by confining data processing to the physical device. While this architecture does not eliminate every risk, it shifts data governance back to the user end. Opera has leveraged this privacy-first positioning as a key competitive advantage across its product lines for years.
The Hardware Bottleneck and Performance Predictability
Shifting compute workloads to the client side introduces explicit trade-offs. Running models locally increases local resource consumption, directly impacting system performance. Opera acknowledges that execution speed and reliability will depend entirely on the consumer's hardware configuration.
While eliminating server expenses solves the software vendor's margin problem, it ties application performance directly to fragmented consumer hardware. Companies adopting the standard trade predictable cloud latency and centralized compute control for variable client-side execution speeds across diverse user devices.
Opera attempts to bypass the recurring API cost structure by offloading execution entirely onto the user's processor and memory.
What it means
Who actually foots the bill for running local AI models? Opera neatly zeroes out its own cloud infrastructure and API expenses by shifting text generation and semantic analysis straight onto users' machines under the banner of data privacy. Software makers secure zero marginal compute costs precisely because Mac and PC owners pick up the slack, spending their own hardware resources to run open-weight systems like Google Gemma. Opera's cloud savings simply turn into higher local hardware load and electricity bills for its users.