Tethers QVAC II Launches an Unprecedented Expansion of AI Training Dataset Translation: Tethers QVAC II Launches an Unprecedented Expansion of AI Training Dataset

Tether Data’s AI division, QVAC, has notably enhanced the «largest publicly accessible synthetic dataset» for artificial intelligence training.

QVAC Genesis II introduces an additional 107 billion tokens, bringing the total to 148 billion across 19 educational domains. This expansion «significantly enhances» the scale, depth, and quality of reasoning.

The second version builds upon the foundation laid by the first. It encompasses 10 new fields, including chemistry, computer science, statistics, machine learning, astronomy, geography, econometrics, and electrical engineering.

QVAC Genesis II replicates university-level physics and, alongside Genesis I, forms «the most comprehensive synthetic educational dataset ever released to the public.»

At the core of this release is a new information generation approach called Option-Level Reasoning, designed to extract structured reasoning from model errors and correct answers.

«The result is a dataset that emphasizes clarity, causal relationships, and decision-making, rather than mere surface correctness,» states the company’s blog.

Tether emphasized that QVAC is focused on training the model to think, reason, and explain, rather than just imitate.

«Currently, most programs are optimized for fluency rather than understanding. With this release, we are moving beyond volume to address structure, reasoning, and clarity,» said the firm’s CEO, Paolo Ardoino.

In May, Tether announced a new platform, QVAC, aimed at developing «infinite and ubiquitous intelligence,» which envisions the «launch and evolution» of AI agents on user devices rather than in the data centers of large companies.

In June, Ardoino predicted that within 15 years, a trillion AI agents utilizing Bitcoin and USDT for transactions would emerge.