According to an official blog post, QVAC, the AI research division of Tether Data, has launched the QVAC Genesis initiative and released Genesis I, a synthetic dataset. This dataset, containing 41 billion text tokens, will help build smarter and more accurate STEM language models globally. These models can be trained to understand words and their associated logic. Rigorously validated against educational and scientific benchmarks, it demonstrates superior reasoning and problem-solving capabilities in subjects like mathematics and physics. This is the first publicly available synthetic dataset purpose-built and rigorously validated for educational content, addressing the shortage of public training datasets in key STEM fields. QVAC Genesis I aims to return AI training power to the public through open, high-quality data.

Tether Data also released its first consumer app, QVAC Workbench, for AI enthusiasts and others. It supports a variety of large-scale language models and other tools. The app is compatible with smartphones (currently Android only, with iOS coming soon) and desktop platforms, offering comprehensive native device support. Chat and interaction data is 100% private, and the "delegated inference" feature connects the mobile and desktop versions to fully utilize workstation resources.