Tether has unveiled the open source QVAC Genesis I dataset and QVAC Workbench to decentralise AI development, giving researchers and developers open access to scientific training data and local AI execution tools.
Tether Data has launched the open source QVAC Genesis I dataset, a significant step toward decentralising AI development and democratising access to high-quality training data. Designed to enhance AI reasoning and problem-solving in STEM fields such as mathematics, physics, and medicine, the dataset comprises 41 billion synthetically generated tokens focused on strengthening AI’s logical reasoning and understanding of cause-and-effect relationships. This release directly challenges Big Tech’s control over AI training resources.
“Most AI today sounds smart but doesn’t truly think. We designed this dataset to help models understand cause and effect,” said Paolo Ardoino, CEO of Tether.
The QVAC Genesis I dataset is structured to improve AI’s capacity for critical thinking and scientific reasoning, converting high-quality educational content into structured learning data. It addresses the limitations of current open-source models, which often struggle with complex problem-solving tasks, enabling developers and researchers to train models capable of deeper analytical reasoning beyond pattern recognition.
Complementing this release, Tether has introduced QVAC Workbench, a local AI tool that enables users to run AI models directly on their own devices across Android, Windows, macOS, and Linux, with iOS support to follow. The platform supports open-source models such as Llama, Medgemma, and Qwen, and features Delegated Inference, which connects mobile devices to more powerful desktops for improved performance.
Tether’s initiative reflects a broader mission to reduce dependence on centralised, corporate-controlled AI platforms. As Ardoino noted, “Intelligence shouldn’t be centralised,” underscoring Tether’s vision for peer-to-peer, transparent, and open AI systems accessible to all.
