Tether Data today announced the launch of QVAC Fabric LLM, a new comprehensive LLM inference runtime and fine-tuning framework that makes it possible to execute, train and personalize large language models directly on everyday hardware, including consumer GPUs, laptops, and even smartphones. What once required high-end cloud servers or specialized NVIDIA systems can now happen locally on devices people already own.
Shifting the Paradigm of LLM Development
High Performance LLM Inference Runtime and Fine-tuning have traditionally been reserved for companies with access to expensive infrastructure, but QVAC Fabric shifts that model entirely. It is the first unified, portable, cross-platform, highly-scalable system capable of full LLM inference execution, LoRA, and instruction-tuning across mobile operating systems (iOS and Android) as well as all other laptop, desktop, and server environments (Windows, macOS, Linux). This allows developers and organizations to build, deploy, execute, and customize AI privately and independently, without cloud dependency, vendor lock-in, or sensitive data leaving the device.
A major breakthrough in this release is the ability to fine-tune models on mobile GPUs such as Qualcomm Adreno and ARM Mali. This is the first time a production-ready framework has enabled modern LLM training on smartphone-class hardware. It opens the door to personalized AI that can learn directly from users on their devices, preserving privacy and functioning even without an internet connection, and powering a new generation of highly resilient, anti-fragile, on-device AI applications.
QVAC Fabric LLM also expands the capabilities of the llama.cpp ecosystem by adding fine-tuning support for modern models such as LLama3, Qwen3, and Gemma3. These models, previously unsupported in this environment, can now be fine-tuned through a simple, consistent workflow across all hardware types.
By enabling training across a wide range of GPUs, including those from AMD, Intel, NVIDIA, Apple Silicon, and mobile chips, QVAC Fabric LLM breaks the longstanding assumption that meaningful AI development requires access to specialized, single-vendor hardware. Consumer GPUs now play on equal footing, and even mobile devices enter the conversation as legitimate training platforms. This is a significant step toward diversifying the hardware available for AI development.
For enterprises, the implications extend beyond convenience. Organizations can now fine-tune AI models in-house, on secure hardware, without exposing sensitive data to external cloud providers. This makes it easier to meet privacy, regulatory, and cost requirements while still deploying modern AI tailored to internal needs. It moves fine-tuning from centralized GPU clusters to the broader device fleet companies already manage.
Paolo Ardoino, CEO of Tether, said the release demonstrates the company’s commitment to making AI more accessible and more resilient. “AI should not be something controlled only by large cloud platforms,” Ardoino said. “QVAC Fabric LLM gives people and companies the ability to execute inference and fine-tune powerful models on their own terms, on their own hardware, with full control of their data. This is the future of privacy-preserving, decentralized, hyper-scalable, and ubiquitous AI.”
Tether Data has released QVAC Fabric LLM as open-source software under the Apache 2.0 license, along with multi-platform binaries and ready-to-use adapters on Hugging Face. Developers can begin fine-tuning with only a few commands, lowering the barrier to AI customization in a way that has not been possible before.
QVAC Fabric LLM represents a practical shift toward decentralized, user-controlled AI. While much of the industry continues to focus on cloud-first solutions, Tether Data is making advanced personalization accessible on local edge hardware to ensure operational continuity in high-latency geographical areas (i.e., emerging markets), to provide an anti-fragile, privacy-first, highly-resilient, and scalable AI platform.
Developers can evaluate QVAC Fabric LLM on Hugging Face via the published technical overview available here.
About Tether Data
Tether Data, S.A. de C.V. (“Tether Data”) is part of Tether’s broader vision to advance freedom, transparency, and innovation through technology. Its mission is to enable people and organizations to connect and share information directly, without unnecessary intermediaries. By creating secure, peer-to-peer systems, Tether Data gives users greater control over their data, communications, and digital interactions. Tether Data aims to redefine how information flows across networks by replacing centralized models with decentralized infrastructure designed for privacy, efficiency, and resilience. The company’s goal is to make global connectivity faster, safer, and more private, empowering individuals and institutions alike to exchange information freely and securely.
About QVAC
QVAC is Tether Data’s advanced AI research initiative dedicated to building open, decentralized, and adaptive intelligence systems. Its mission, "Local AI. Infinite Intelligence. No Compromise," envisions a world where AI lives and learns on any device, empowering individuals and communities rather than concentrating power in corporate data centers.