At All Costs

Artificial intelligence can change the lives of billions of people. We believe it only will if everyone can use it: privately, at high performance, and without restriction.

Today that is not the case. The most capable systems are closely held by a few closed-source labs in the United States, while a growing number of open-weight models come primarily from China. If the world keeps marching forward as it is, with those labs gaining more political and economic influence through their capital structures, it is not far-fetched to imagine a world where they decide what intelligence may be used for, and monetize it for their own benefit. The same intelligence could produce resilient crops that feed millions, treatments for diseases we have never been able to cure, or social media algorithms built to capture as much of our attention as possible. Someone decides which. Right now, it is not you.

Our goal is to give that choice back to you, by distributing model capabilities as broadly as possible while enabling trust for every end user: anonymous by default, unrestricted by design, high performance by demand. We believe AI will bring about an age of abundance, not catastrophe. Its benefits should reach everyone and not just those who control access to it.

Silares is a team of algorithm developers, performance engineers, and reverse engineers. We build ultra-low-latency systems for frontier reasoning models, and will make them available to everyone, privately. In this, we are grateful to be backed by angel and institutional investors who share our commitment.

In the coming months, we will begin releasing our work to the public: research on the systems that make our models fast, tools for developers to integrate frontier models into their own products, and applications for anyone who wants to use AI privately. We look forward to working with partner companies, developers, and consumers along the way, and to learning from the people who use what we release.

It will not be easy, but our execution will set the benchmark as we build toward a future where intelligence belongs to everyone. At all costs.