Tether releases 460M-parameter VisionPsy-Nano models for on-device vision AI
QVAC published Apache 2.0 weights, phone benchmarks and inference code, while its headline scores remain based on in-house evaluations.
By Ryan Merket · Published
Primary source: QVAC on X
Why it matters
Tether is using QVAC to expand from stablecoins into edge AI infrastructure. A capable 460M-parameter VLM could lower cloud inference costs and keep sensitive images on users' devices, provided independent testing confirms QVAC's results.
Tether Data's QVAC research group released two 460M-parameter vision-language models on July 29th, giving developers open weights and mobile inference code for applications that need to interpret images without sending them to a cloud service. QVAC announced VisionPsy-Nano in a thread on X and published the model collection under the Apache 2.0 license.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.