Skip to message box
Virgo AI by KSN
Explore tools
Pick a tool, then pick its model: Virgo server models work in any browser, and on-device models run privately on yours.
Speech to text
Record your voice or upload audio and get a transcript with Whisper.
Text to speech
Read text aloud with a natural server voice, the Khmer voice or your device's voices.
Image
Describe a picture and generate it on your device.
Server models work in any browser. Janus-Pro runs on your device: it needs WebGPU and downloads about 2 GB once.
Video
AI images from your prompt, animated into a short clip with pan, zoom and crossfade.
Each frame is one AI image: a few seconds on the server, about a minute on this device.
Music
Describe a sound and compose a short clip with MusicGen.
The first song downloads a model (about 600 MB), then it is cached.