Today we launch QVAC SDK - Local AI SDK that works on all mobile and desktop platforms. Build once, run everyhwre! Features both on-device inference and finetuning.

Tether AI breakthrough Tether AI team just released new version of QVAC Fabric to include the World’s First Cross-Platform BitNet LoRA Framework to Enable Billion-Parameter AI Training and Inference on Consumer GPUs and Smartphones. Background Microsoft's BitNet uses one bit architecture to dramatically compress models. Traditional LLMs operate on full-precision computation, where weights are stored as complex, high-resolution numbers. The innovation of BitNet is that it shrinks these weights into a tiny ternary range of only -1, 0, and 1. significantly reducing memory usage and computation. LoRA, is a parameter-efficient fine-tuning technique that reduces the number of trainable parameters by up to ninety-nine percent. Together they slash memory and compute requirements. Yet BitNet has mostly been limited to CPU or CUDA NVIDIA backends, and lacked the support of LoRA fine-tuning. Enters QVAC Fabric: the unlock Today, with QVAC Fabric LLM, is the first time BitNet LoRA fine-tuning and inference work cross-platform across GPU vendors and operating systems using Vulkan and Metal backends. That means support for AMD, Intel, Apple Metal and also Mobile GPUs. And for the first time ever, BitNet inference runs efficiently on smartphones using mobile GPUs. On flagship devices, GPU inference is 2 to 11 times faster than CPU while using up to 90% less memory than the full precision models. The biggest unlock: QVAC Fabric LLM support for BitNet LoRA fine-tuning on heterogeneous GPUs. Our team was able to demonstrate this by fine tuning models up to 3.8 billion parameters on all flagships phones such as Pixel 9, S25 and iPhone 16 and up to 13 billion parameter models on the iPhone 16. Github repositories: github.com/tetherto/qvac-fab… : general QVAC Fabric codebase github.com/tetherto/qvac-rnd… : specific QVAC Fabric's BitNet knowledge base, architecture docs and pre-built binaries What does it mean? What used to require dedicated GPUs now runs on consumer hardware. This breakthrough is the first real-world signal of a local private AI that can truly serve the people. And this is just the beginning. In the next months and years Tether will relentlessly continue to invest significant amounts of resources and capital to continue to research and develop open-source intelligence that can scale and evolve on local devices, providing maximum utility and privacy to its users. The era of Stable Intelligence has just begun. Free as in freedom.

We're happy to announce a long-term partnership with Motorola. We're collaborating on future devices meeting our privacy and security standards with official GrapheneOS support. motorolanews.com/motorola-th…

Introducing Scudo. A new way to measure the value of gold on-chain. Scudo is a simple, intuitive unit that makes Tether Gold ( XAU₮) easier to use, track, and transact. 1 Scudo = 1/1000 of an XAU₮ (Gold Ounce), giving you a practical and accessible way to send and receive gold in smaller amounts.

Big thanks to @luganomycity and @pow_space for hosting the QVAC team during AI Week 🇨🇭 We’re proud to share our vision for a local-first AI platform — built for privacy, performance, and full user control. #QVAC #AIWeekLugano @LuganoPlanB

EDGE/ON-DEVICE AI INFERENCE AND FINE-TUNING IS HERE. Tether Data just released QVAC Fabric LLM, and it creates a new foundation for how AI is built and deployed. It is the world's first Edge-First Inference Runtime & Fine-Tuning Framework. Here’s the simple breakdown:👇 The Old Way: To run AI models (Inference), you needed the cloud and pay a subscription. To evolve and customize AI (Fine-Tuning), you needed expensive clusters. • It was centralized. • It was reserved to the elite. • Your data had to leave your device. The QVAC Fabric Way: We built a unified, cross-platform system to EXECUTE (INFERENCE) and PERSONALIZE (FINE-TUNE) models on the hardware you already own. • Laptops (Windows, Mac, Linux)? Yes. • Consumer GPUs? Yes. • iOS & ANDROID SMARTPHONES? YES. How we did it (The Geeky Part): We extended the llama.cpp engine to add more function instrumentation, generalized support for new models and introducing state-of-the-art, highly extensible LoRA fine-tuning capabilities. We made it cross-platform and vendor-agnostic. It’s highly efficient, meaning it doesn't need a nuclear reactor to run—just your device battery. Why is this a big deal? For Developers: You don’t need a massive budget or a cloud provider to build custom AI anymore. You can build, test, and fine-tune models like Llama 3 or Gemma 3 directly on your MacBook, Linux rig, Windows desktop or even your mobile device. It’s open-source and uses llama.cpp, so it’s super lightweight. For Regular People: Imagine an AI assistant that actually learns from you—your notes, your style, your preferences—but none of that data ever leaves your phone. It lives locally but can scale infinitely. It learns locally. It can work offline. It’s truly your AI, not a corporate rental. The only solution that can truly serve anyone, including the billions of people that can’t afford big-tech expensive subscriptions. The TL;DR: We moved the entire AI lifecycle—execution and evolution—from the cloud to people’s devices. No vendor lock-in. No spying. Just pure, ubiquitous intelligence. Open Source.Multi-platform binaries. Ready today. QVAC - Your Device. Your AI 🔗 Read the QVAC Fabric-LLM Tech Overview & Get the Code: huggingface.co/blog/qvac/fab… hashtag#QVAC hashtag#LocalAI hashtag#Llama

Today, we released a base model for Genesis I that allows AI researchers to replicate published results instantly, enhancing scientific transparency and reproducibility. It also makes it far easier for researchers and practitioners to test, compare, and build on Genesis I without needing to train from scratch using large high-end GPU resources. Genesis I provides a practical foundation for developing next-generation STEM learning assistants that genuinely understand complex STEM concepts. Our upcoming Genesis II synthetic dataset plans to cover more STEM topics to broaden the coverage and progress towards full completeness. #QVAC Read more: Blog post huggingface.co/blog/qvac/gen… Try the model huggingface.co/qvac/genesis-… x.com/Tether_to/status/19817…

Follow @qvac and watch the velocity at which this is getting shipped by @paoloardoino @pingsns and the whole team. On-Device AI (private & p2p) is the future.

QVAC first demo (android mobile) of its local on-device fine-tuning generalized framework, which aims to bring LORA fine-tuning to any device, scaling from the smallest embedded device to any smartphone, laptop, server, operating system. In combination with WIP multi-node features, we expect QVAC to become the standard for local on-device single/multi node inference and fine-tuning. Imagine a horde of AI agents being able to learn from your emails, notes and messages, maintaining 100℅ privacy, leveraging only your own devices. A P2P private intelligence network at your fingertips. That's how QVAC personal assistant will look like. Future of Intelligence won't be centralized. Infinite Ingelligence vibes.

In Lugano, we use Keet ;) x.com/callebtc/status/198175…

WTF @x @nikitabier? Why is this news not in my feed? I only found out about this because I heard it from David Bennett on the Bitcoin And podcast. @grok what is going on with the algo? I thought they were using AI to fix the feed. I have gotten two straight days of people sperging out about EBT/SNAP payments, but not this sort of important news.

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. 🤖🤖

Modular, self-custodial infrastructure for the next generation of wallets. Built for humans and agents. Native support for USDT0 XAUt0 to accelerate payments on Polygon. WDK by @tether is here and open-source.

Today Tether introduces WDK, the Wallet Development Kit, one of the core pillars of our Stable Company strategy. At Tether, we imagine a world where humans, autonomous machines, and AI Agents have the freedom to control their own finances and transact without boundaries. In that future, where programmable money is the social network across the fabric of society, WDK by Tether is the cornerstone of the financial backbone, a fully open-source, self-custodial toolkit designed to be modular, independent, resilient, and infinitely scalable, enabling trillions of wallets. Self-custodial wallets are the cornerstone of a free and resilient monetary infrastructure. If we continue to build towards this common goal, we'll be Unstoppable Together. For a detailed overview of WDK: 1. watch the official video 👇 2. read the announcement 3. read the docs 4. checkout the codebase 5. contribute back

It’s time to build ⚡️ Today, together with the biggest names in the space, we’re launching a global developer challenge to bring bitcoin to the open-source apps used by millions every day. 💰 $25k prize pool 🌍 50 communities 🚀 ∞ possibilities 🧵 time2build.dev

Made for America

The team at @qvac is working incredibly hard to bring fully local, private, p2p native AI to everyone on the planet. Very proud to be part of it. It's an incredible feat, and there are many challenges ahead. But, as they say: "the best way to predict the future is to create it".

Marco @elchiapp, Head of Product at QVAC Health, explains the last private beta app update. Still lot of work in progress. Current focus is building all the underlying structured local datastore so that anything can be built on top of it. Enjoy 💪

Why QVAC @tether ?

Thousands joined for an evening to travel beyond time and space and basked in bliss, against the backdrop of the ancient amphitheatre in Plovdiv, Bulgaria, the oldest city in Europe.