Has typical LLM quirks, but none too glaring, hardly just a Gemini problem.
Common phrase is “gesturing” with hands. And Gemini Pro really loves using ellipses (though it’s decreased in the June version). I can regex them away. But logit bias would be a nice feature for some of the “LLM-isms.”
More robust documentation for all Google models would be a big plus. A lot of people I talk to find the docs lacking.
Overall, well done. Appreciate the action on user feedback, especially giving some love to creative writing. It would be an easy corner to neglect. 10/10 version update in my book. Release this