document upload and extraction flow should be faster in locallab now. i ran some evals and gemma 4 26b is now about 2.5x faster than it used to be for that flow on my machine. also, i plan to make a desktop app to make it easier for people to run the app. it will still require you to be running ollama though. thinking about allowing users to use @PayPerQ with private only models instead of ollama. that way you won't accidentally send your lab results to scam altman.



https://github.com/SamSamskies/locallab