PDF Reading in Grok is great now It can Read Very Long PDFs, can see Visual data inside PDF and it's very accurate
The sad part of this is I tried this "feature" a month ago and it was a miserable failure. Except for my following you on X, I would have absolutely no way to know that this feature now actually works, and works well. Mix that with all the feature announcements that neglect to cite a platform, and it makes being the alpha testers, that we are, much worse than need be.
I just spent an hour with grok.com and this improvement is there as well. Until this fix, Grok was terrible at reading pdfs, even for text. They were being rendered to images then OCRed. Typically half or more of the text was lost. Tonight Grok got an "A" for reading my electric utility bill and rendering all of the 50-60 line item details to a csv file. The whole Grok.com project experience is poor and buggy, but the immense potential is obvious. Thank you xAI geniuses.