Gemini 3.5 Flash is the model most Google API traffic runs through — not the flagship, but the tier that has to be fast, cheap and good enough. It carries a 1-million-token input window, returns up to 65,000 tokens, and reads text, images, audio, video and documents. The distinguishing feature is a thinking dial with four settings — minimal, low, medium and high — where medium became the default, a quiet admission that the previous generation was spending reasoning tokens on questions that did not need them. Google positions it for sustained agentic and coding work rather than single-shot answers, and supports thought preservation across turns so a long agent run does not restart its reasoning at every step. It is available on the free tier and priced at 1.50 dollars per million input tokens and 9.00 dollars per million output tokens on the paid tier. Note the knowledge cutoff: January 2025, older than the model's own release year.