ai input here put your google api here ai output here Gemini 3.5 Flash-Lite (fastest and with least token usage) Gemini 3.1 Pro preview (best model but will eat through your tokens) gemini-3.7-flash (mid range model slightly dumber but still useful)