TechFlow News: On May 7, the Tether AI Research Group launched the QVAC MedPsy series of medical language models, designed to run locally on low-compute devices such as smartphones and wearables—reducing reliance on cloud infrastructure. According to official reports, the 1.7B-parameter version achieves an average score of 62.62 across seven closed-domain medical benchmark tests, outperforming Google’s MedGemma-1.5-4B-it; the 4B-parameter version attains an average score of 70.54, surpassing even larger models including MedGemma-27B-text. Tether states that the model also lowers inference costs and has released quantized GGUF versions optimized for local deployment.
Add to Favorites
Share to Social Media
