Deep Tide TechFlow News, May 7th, Tether AI Research Group launched the medical language model series QVAC MedPsy, primarily designed to run locally on smartphones, wearables, and other low-compute-end devices, reducing reliance on cloud infrastructure. The official stated that the 1.7 billion parameter version scored an average of 62.62 across 7 closed medical benchmarks, surpassing Google's MedGemma-1.5-4B-it; the 4 billion version scored an average of 70.54, outperforming larger models including MedGemma-27B-text. Tether indicated that the model can also reduce inference costs and has provided a quantized GGUF version suitable for local deployment.