Google's image model is among the strongest you can use for nothing, built into the Gemini app and quick to iterate with in plain language. It handles people and scenes well, and being part of Gemini it sits a sentence away from the rest of your assistant.
Content filters are stricter than the open tools, so it declines more prompts, and the very newest tricks can trail the paid leaders by a release. As a no-cost starting point it is hard to beat.
