Daum I must try the CLI version too. But the biggest disadvantage is always have to work with keyboad using multiple key strokes which could be done with 1 mouse click
Lol… you forgot about the F5 and just speak… even further press Shift+Tab to cycle trough plan , execute modes :)))) … them simply read the Artifact by /artifacts and approve, reject etc.. you guys love too much clicking and the mouse … but don get inside the Desktop UI apps that will never bring 100% of functionality, couse there is no place or it becomes too clumbsy…
If simple / slash is too much to see entire tree is too much… then stay to UI instead TUI.
Yeah, fair point about autocomplete and command trees, but that still misses the actual bottleneck.
TUI and voice work fine for triggering simple prompts, but they completely break down for heavy spatial tasks. When an update drops 2,700 lines across 15 files, no terminal list or voice command is going to replace a side-by-side visual diff. Trying to “listen” to code or cycle through terminal text for complex reviews isn’t efficient it’s just painful
More importantly, telling everyone to “just switch to CLI” is a total cop out. Pointing out that 2.6.0 completely broke core UI flows like the Review button is valid criticism. A working terminal fallback doesn’t excuse shipping a messy, half-broken desktop experience to begin with
Earlier, we could see the reasoning process for Gemini models, which made it easy to follow the model’s approach and catch logic issues early. With the new update hiding the thinking steps, we’ve lost that transparency and have no visibility into how the model is approaching the problem.
Exactly! It was visible while the model is working. But not it’s optional.
Even though it’s not a big deal since we still have the option to expand to see the thinking text, we all can understand your point that it was better before the new update.