It's crazy to me that there's no screenshots on the blog post. And when you click through to download it there's a bunch of aspirational pictures, but only one short clip showing what it actually looks like. Why? Is it too hard to show a screenshot?

I still can't stand the fact that in order to use their mobile app you must allow sharing data for training. No thanks will close the app and it seems like they follow the same thing with their desktop app.

Outside of that, if I want to use Gemini through the browser they link allowing your data to be trained on with persistence of chats. If I don't want my data trained on I must allow them to train on my data.

For those reasons it's pretty much unusable for me.

Is it still gemini 3.6? It's crazy that i still have 3.6 and if you want 3.8 you need to use API.

About time, I've been having resource usage and stuttering issues on browser from time to time (both on Firefox and Chromium). It seems a little bit better on desktop. I still don't understand why aren't they merging Antigravity - Gemini - Spark under one app though. Antigravity is still looks confusing to me.

My biggest bugbear with Gemini is that I can’t organise chats into folders and projects.

It’s such a powerful model, truly useful for general purpose and marketing tasks… but the UX is basic beyond belief.

I literally find myself using AI Studio for general tasks as it’s a better workflow.

Sorry, but nobody really cares. In the objective sense, indeed. There are so many players already in this area, and Google's products in this area do not stand out with anything particularly impressive (at this moment, perhaps if they released in 2023 we'd be having different conversation).

Saying this as s.o. calling their APIs on a daily basis for work, but would never let their models touch my everyday OS.

Just wanted to say that if you want to learn any hard topic, Gemini > ChatGPT. I pay for both. Don't yet know about Claude.

Interesting. I wonder if this will evolve into something closer to Codex or Claude Code, where it can actually work across files and tasks on your computer, not just act as a desktop chatbot.