Testing VOCR with a local model for computer use! Qwen3.8-Flash-Next is the first local model small enough to run on my M3 Max with 64 GB that could tackle my UI challenge.
More stats on the video description, but it's pretty slow. I edited out all the waiting for the model to respond for demo purposes. It actually took about 39 minutes from start to finish.
I'm pretty hopeful about potential and possibilities for #accessibility with #AI!
https://youtu.be/HYWOVqt5km0
@vick21 @ZBennoui @FreakyFwoof