Tether Launches New AI Framework: Phones Can Train Billion-Parameter BitNet Models

Stablecoin issuer Tether recently announced that its AI platform QVAC Fabric has launched the world's first cross-platform LoRA fine-tuning framework for Microsoft BitNet (1-bit LLM), enabling training of billion-parameter language models on ordinary hardware like laptops, consumer GPUs, and smartphones.

Test results show that a 1 billion-parameter BitNet model takes about 1 hour and 18 minutes to fine-tune on Samsung S25, and about 1 hour and 45 minutes on iPhone 16; the team even successfully fine-tuned a 13 billion-parameter model on iPhone 16. The framework significantly lowers memory and computing power thresholds, supporting various chip types.

This technology aims to reduce reliance on large-scale cloud computing, promote decentralized AI and federated learning, allowing model training to be done on local devices and protecting data privacy.