pi-vision

blueye-y

esm

Give text-only pi models vision — describe images with a vision model you pick via an interactive picker, then hand off the text description to non-vision models. Includes direct clipboard paste (image-paste integration).