Hi Google/Gemini team,
I’m building GainScan AI, a fitness and nutrition app using Lovable.dev, Supabase and the Gemini API.
I’d like some guidance on:
1. What would be the approximate Gemini API cost for one AI Coach message and one meal-photo analysis, including image/input and output usage?
2. Which current Gemini model would you recommend for accurate food identification, portion estimation, calories and macros while keeping costs reasonable?
3. What is the best production approach to improve meal-scan accuracy and represent uncertainty/ranges correctly?
4. What is the recommended architecture for securely calling Gemini from a Lovable/Supabase app without exposing the API key?
5. Most importantly, my app is designed for users aged 13+. Does the current Gemini API policy allow an API-powered fitness/nutrition app to be accessible to users aged 13–17? If not, is there an approved way to structure this?
I would really appreciate an official clarification, especially regarding the age requirement and API usage costs.
Thank you.