Listen to the article
Tether Data, the technology arm of the world’s largest stablecoin issuer, is expanding into artificial intelligence with the launch of what it calls the world’s largest synthetic dataset for STEM-focused AI models.
Earlier today, the company unveiled QVAC Genesis I, a 41-billion-token dataset built to train science- and engineering-oriented language models, and QVAC Workbench, a cross-platform local AI application that runs models directly on consumer devices. Per Tether, QVAC stands for “QuantumVerse Automatic Computer.”
“QVAC is Tether’s answer to centralized AI. An entirely new paradigm where intelligence runs privately, locally, and without permission,” the QVAC website mission statement said. “No clouds. No gatekeepers. Just you, your machines, and unstoppable intelligence.”
The move marks a striking escalation of Tether’s ambitions beyond finance. The company said the dataset had been validated on math, physics, biology, and medical benchmarks, and was designed to “level the playing field” for open-source AI by giving researchers an alternative to proprietary data controlled by companies such as OpenAI and Google.
Tether’s AI first launch.
Local on-device AI is evolving.– QVAC Workbench, mobile/desktop app for using and experimenting many AI models locally on device with 100% privacy
– QVAC Genesis I, largest synthetic pre-training datasets for Large Language Models (LLMs) to date.… https://t.co/79lYhsobuc— Paolo Ardoino 🤖 (@paoloardoino) October 24, 2025
While QVAC Genesis I itself isn’t a financial product, the broader QVAC ecosystem is being built with clear links to Tether’s crypto infrastructure. In earlier company statements, Tether said QVAC’s architecture will eventually integrate Bitcoin and its own stablecoin, USDT, enabling AI agents to transact autonomously using digital assets.
This suggests the initiative could evolve beyond data and local AI tools into a network where intelligent agents can not only learn and reason, but also pay, trade, and interact directly through blockchain rails.
“Intelligence shouldn’t be centralized,” said Paolo Ardoino, Tether’s chief executive, in a statement accompanying the release. “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.”
QVAC’s consumer app
The company also released a free consumer-facing app called QVAC Workbench for smartphones—Android for now, and iOS “within a few days”—as well as desktop platforms (Windows, macOS, and Linux). “With QVAC Workbench, all chats and interactions with the AI models remain local on-device, where data is owned by the user and remains 100% private,” the company said.
It also introduces a peer-to-peer feature called delegated inference, which allows the mobile app to offload heavy computation to a desktop workstation while keeping all data private and local.
Unlike conventional training material scraped from the public internet, QVAC’s dataset is fully synthetic: generated, filtered, and validated by models trained on educational and scientific materials. Tether claims the data enables models to “reason, solve problems, and think critically” rather than simply mimic text patterns. The full technical breakdown is available in the QVAC research blog.
The QVAC Workbench app lets users run large language models such as Llama, MedGemma, and Qwen entirely on their phones or computers.
Tether framed the twin releases as part of a larger effort to create “local intelligence,” or AI that operates independently of cloud servers. The company, which already dominates the stablecoin market with its USDT token, is positioning its AI unit, Tether Data, as an advocate for decentralized infrastructure that keeps both money and information outside corporate control.
Who owns your AI?
The project arrives amid intensifying debate over synthetic data’s role in model training. While it promises privacy and scalability, skeptics caution that synthetic training data can amplify the biases or errors of their parent models, potentially locking in distorted reasoning patterns. Tether’s announcement did not specify which generative systems produced Genesis I’s content or how its quality assurance was performed.
Even so, QVAC Genesis I represents one of the boldest open-data experiments yet from a private crypto firm. If Tether’s claims hold up under scrutiny, then it could give independent researchers and smaller labs a new foothold in the AI race, and signal Tether’s determination to influence not only the future of finance, but of artificial intelligence itself.
