Tether Data’s AI research division, QuantumVerse Automatic Computer (QVAC), has released QVAC Genesis I, a large-scale synthetic dataset designed for advanced AI training and language models, especially those focused on STEM domains.

Built from 41 billion text tokens, QVAC Genesis I stands as the most extensive synthetic dataset ever created for AI training. It’s purpose-built to support the creation of language models that can reason, analyze, and solve complex problems in scientific domains such as mathematics, physics, biology, and medicine.

In addition to the giant dataset, Tether’s AI research team introduced a consumer app for local on-device AI processing, called QVAC Workbench.

The app supports various AI models, including Llama, Medgemma, Qwen, SmolVLM, and Whisper, and is available on Android devices, with iOS compatibility coming soon. Desktop versions for Windows, macOS, and Linux are also available.

QVAC Workbench allows users to maintain privacy by keeping all AI interactions local on their devices, with a “Delegated Inference” feature enabling peer-to-peer connections between mobile and desktop applications.

Tether CEO Paolo Ardoino said in a statement that the company’s latest AI initiatives embody Tether’s mission to make intelligence as decentralized and open as information.

“Intelligence shouldn’t be centralized,” said Paolo Ardoino, CEO of Tether. “With QVAC Workbench and Genesis I, we’re opening the door to infinite intelligence, AI that lives, learns, and evolves locally on your own device. We believe that intelligence, like information, should be free, accessible, and owned by everyone, not locked behind corporate firewalls or sold as a service.”

The QVAC Genesis dataset has been validated across educational and scientific benchmarks and represents the first publicly available synthetic dataset specifically built for education-specific content.

“Whether it’s a phone, a robot, or a wearable, intelligence s

Go to Source to See Full Article
Author: Vivian Nguyen