Tether Launches QVAC Fabric Framework for LLM Fine-Tuning on Smartphones

Stablecoin issuer Tether is expanding into AI with the launch of QVAC Fabric, an infrastructure framework that combines the low-bit BitNet model with the parameter-efficient fine-tuning technique LoRA. The framework aims to reduce large language models' reliance on costly data-center GPUs, break the concentration of computing power among technology giants and enable personal devices to participate in AI training.

Tether most recently announced that QVAC Fabric can run LLM fine-tuning and training on smartphones, including iPhones and Samsung devices, as well as consumer-grade GPUs. Users can process data locally instead of uploading all of it to the cloud. Tether did not disclose an exact release date, a list of supported devices, model parameter sizes, training speeds or commercial pricing. Its real-world performance has yet to be independently tested and verified.