Tether представляет QVAC Genesis II: революция в обучении ИИ с помощью крупнейшего синтетического набора данных Translation: Tether unveils QVAC Genesis II: a revolution in AI training with the world’s largest synthetic dataset.

The AI division of Tether Data, known as QVAC, has significantly expanded the «largest publicly available synthetic dataset» for artificial intelligence training.

QVAC Genesis II introduces 107 billion new tokens, bringing the total to 148 billion across 19 educational fields. This «substantially enhances» the scale, depth, and quality of reasoning capabilities.

The second version builds upon the foundation laid by its predecessor and incorporates 10 new disciplines, including chemistry, computer science, statistics, machine learning, astronomy, geography, econometrics, and electrical engineering.

QVAC Genesis II reconstructs university-level physics, forming with Genesis I the «most comprehensive synthetic educational dataset ever presented to the public.»

At the core of this release is a novel approach to information generation known as Option-Level Reasoning. This method is designed to extract structured reasoning from both model errors and correct answers.

«The result is training data that emphasizes clarity, causality, and decision-making rather than mere superficial correctness,» the company’s blog notes.

Tether emphasized that QVAC aims to train models to think, reason, and explain, rather than just mimic.

«Today, most programs are optimized for fluency of speech rather than understanding. With this release, we move beyond mere volume into structure, reasoning, and clarity,» remarked the firm’s CEO, Paolo Ardoino.

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

In June, Ardoino projected that within 15 years, there will be a trillion AI agents using Bitcoin and USDT for transactions and payments.