A discussion with Gemini about Google AI Plus

Page content

Last night and this morning I have been toying with the idea of the paid tiers and whether they’re worth it. In the proces I told Gemini about my context and it went from advising me the pro tier at 170 CHF per year, to going for the “free” Google AI studio option.

From Gemini to Google AI Studio

I find the pivot interesting because you go from a conversational bot to something more powerful. You go from “How do I do this with this command line tool” to getting python scripts telling you to use the Gemini AI to do things directly rather than passively. It’s the type of pivot that requires the generation of an API key and more.

Tokens

When you have conversations with AI studio it tells you how many tokens you’re using for each question, as well as each file. It’s interesting. The raw cameras used info from my photo library before I cleaned it up was 7.7 million tokens. With a simple grep and uniq command it was reduced to 3200 tokens. Google AI studio has a 1 million token ceiling, and that file was seven times the ceiling. The paradox is that GREP dealt with the issue in milliseconds a day or two ago.

Ask Gemini and Other Tools for Help

With the example above we highlight the importance of asking AI tools to help us reduce their workload. By checking for unique cameras, rather than giving the raw output from a camera search we save 7.1 million tokens.

The API Paywall

Of course, Gemini was encouraging me to go for the pro tier, and then the AI Studio tier in the hope that I would create an app that requires the API key, which I pay per use. In theory I can use the API for free. Although this would be the next step I prefer to ask for a step that allows me to do something locally even if it requires more patience in some cases.

And Finally

After experimenting with Google AI Studio I can see that it is matter of fact, answering the question without the social elements common to Gemini and the wait time is longer as it is using Gemini 3.1 Prop preview, rather than the fast model.

And Finally I was too “chatty” so I reached the rate limit with Google AI Studio. I will explore this in more depth.