Record 098 · Vision · 2024
50%24.98.I
Moondream
vikhyat
“A tiny VLM you can run on a laptop CPU and still caption a photo.”
Moondream made 'what is in this image?' a local, cheap call. HA, Immich, and scripts use it because it is small.
Opened tiny local vision-language.
- ~1–2B VLM
- CPU-friendly captioning
- Apache weights
A vision model that fits in a USB stick's shadow.
