Skip to content
- Tether releases QVAC MedPsy medical AI models for phones, laptops, and servers.
- MedPsy models run without cloud access and support privacy-sensitive healthcare use.
- Tether says MedPsy-4B beats larger MedGemma models on key medical benchmarks.
Tether has released QVAC MedPsy, an open-source medical AI model family for edge devices. The models run on phones, laptops, and servers. They do not need cloud access, according to the company’s official release today.
As per a report, the issuer of USDT said QVAC MedPsy includes 1.7B and 4B parameter text-only models. Both are built for healthcare reasoning on local hardware. Tether said this design supports privacy-sensitive use cases and lowers reliance on remote servers.
Tether Claims MedPsy Beats Larger Medical AI Models
The release claimed that MedPsy-1.7B outperformed Google’s MedGemma-1.5-4B-it by 11.42 points on average. This result included closed-ended medical benchmarks. In addition, Tether claims that the smaller model outperforms larger models on HealthBench and HealthBench Hard.
The company claimed MedPsy-4B also performed well. Despite being significantly smaller, it outperformed MedGemma-27B-text-it on key metrics. Tether was able to claim a 16-point advantage on HealthBench Hard and higher scores on MedXpertQA.
Tether said the models use up to 3.2 times fewer response tokens than similar systems. This could boost the performance on consumer devices. Additionally, the models can be quantized to GGUF, a format that uses less memory without compromising performance.
Also Read: Solana Foundation Says RWA Activity Fuels Institutional Adoption
Tether Expands QVAC With Open-Source AI Models
QVAC MedPsy is licensed Apache 2.0. The models are available on Hugging Face with full GGUF variants. They can be freely used, adapted, and deployed by developers within that framework of permissiveness.
The launch is part of Tether’s QVAC (decentralized AI platform). It also helps the company to go past its stablecoin enterprise. Tether is developing “local-first” tools under the leadership of its CEO Paolo Ardoino that store user data on their own devices.
The release signifies another milestone in the transition to sovereign intelligence, which Tether aims to achieve. The business is integrating privacy, accessibility, and efficiency with medical AI. With its recent action, Tether is going all-in on edge AI, where local processing plays a pivotal role.
Also Read: Chainlink Expands Regulated Crypto Push Through Bermuda Pilot
