The model: Google Veo 3.1, run through Google Flow — text-to-video from one paragraph of prose, dialogue and sound included.
6 tested prompts. Every video below was generated by us in Google Flow, from the exact paragraph of prose on its entry page — no borrowed clips, no scraped prompt lists. Together they cover what people actually ask Veo 3 for: a café conversation with real spoken dialogue, an ASMR glass-fruit slice built around sound, a sneaker product splash, a phone-style UGC ad, a food commercial built on a cheese pull, and a drone flight over a river valley at golden hour. Each entry carries the full prompt text, the model settings, the generation count and the Flow credits it cost — and what broke along the way. These are prompts for Veo 3 with receipts: this page is not a Veo 3 prompt generator, and nothing here is auto-written. For the method behind the set — draft on Lite, finish on Fast, fix one flaw per generation — read our Veo 3 prompt guide.


