The issue with this CLI

[bigbizze]

2025-08-24.01-40-01.mp4

So after having used claude code a lot in the past and testing out gemini. I think I've noticed a few UX differences that might be interpreted by many as worse performance or frustrating interactions.

When Claude code runs, it does two things gemini does not from a UX perspective that stick out to me.

1. It writes the command, tool call or thoughts of the model first, and then results after in two separate updates. It writes a bash command it's executing, and then 500ms later, the result of it. It constantly updates you about what it's thinking rather than saying some variation of "Finalizing the response" 200 times for 5 minutes. This seems minor, but really gives the feeling that something is actually happening.
2. It constantly updates the # of tokens being used by the model for the interaction. Not only does this also reaffirm the feeling that something is happening, but it also makes it easy to tell if the client is really stuck in a loop, or just taking a long time to reply.
3. It shows you the command used for each tool call (It seems like gemini CLI does now?)

These are obviously minor things, but are worth considering imo.