so far, most of the feature work there and it is really generates very good answers and research summaries. I just want to be sure, tier 3, which is ultra also has support of OCR and the ability to choose models while conversing? if so, probably I will add this service to my AI arsenal apps soon.

the voice didn't work for some reason and I didn't understand why. it is working on other apps. will you add more quality voices to the app when hearing the answer and let us really talk and hear the agent in human voice in the future(like we have on chatrgpt for example)