Olares One

The premise behind Olares One is a familiar complaint about cloud AI: recurring subscription fees, privacy exposure when data leaves your premises, and dependence on a network connection. The maker's answer is to move the workload onto hardware you own, running an open-source software stack that it says leaves users in control of their data and applications.

The hardware

• GPU: NVIDIA GeForce RTX™ 5090 Mobile
• CPU: Intel® Core™ Ultra 9 Processor 275HX
• Memory: 96GB RAM

According to the maker, that combination lets the machine run several AI models at once and handle demanding image-generation jobs, with tools such as Stable Diffusion producing high-quality images in a few seconds to a few tens of seconds. The 96GB of memory is presented as enough to run large language models such as Llama 3 70B entirely locally, without the model-size limits of hosted services.

Software and security

Olares One runs what the maker describes as an open-source, multi-layered operating system designed around data ownership, with source code open to audit by users and the developer community. Despite that foundation, the interface is meant to feel like a conventional desktop, with icons, a taskbar and windows, and the maker says Windows applications such as Microsoft Office and Adobe Creative Cloud can run on it. A one-click app development feature uses templates and APIs to let users with little coding experience build their own AI tools. The maker also lists enterprise-grade security features, including data encryption, access controls and a regular update system.

Remote access, gaming and agents

An Olares ID account provides encrypted remote access from phones, tablets or other PCs, so the machine can stay at home while its models and files are reached from elsewhere. The same GPU is pitched for gaming: the maker says Steam games can be played on any screen, including streamed to a tablet or laptop. Beyond personal use, Olares One is framed as a small AI server for building and managing multiple AI agents, for tasks like customer inquiries, inventory management or marketing analysis.

Under the slogan "stop renting the cloud," the campaign claims a 96% cost reduction on the AI stack compared with cloud services, based on owning the hardware outright rather than paying monthly fees.