Skip to main content

Nano Banana

Provider · Google

What Google's Nano Banana models are good at, and how to prompt them for consistent characters and fast iteration.

What is Nano Banana

Nano Banana is the nickname for Google's Gemini image models — Nano Banana, Nano Banana 2 and Nano Banana Pro. They generate and edit images from natural-language prompts, run fast enough for tight iteration loops, and hold a subject's identity across a series of edits better than most alternatives.

They render at 1K, 2K and 4K across seven aspect ratios. There is no quality tier to choose: one pass, one price per image.

What it is good at

  • Character consistency. Keep the same person, mascot or product across a set of shots by passing a reference image with every request.
  • Conversational editing. "Now make it night", "remove the second chair" — successive edits stay anchored to the original frame.
  • Multi-image reference. Blend a subject from one image with the setting or style of another.
  • Fast iteration. Cheap and quick enough to explore ten framings before committing to one.

Why developers reach for it

It is the pragmatic default: low cost per image, no sampler settings, and behavior that stays predictable when you drive it from code. For product photography variants, avatar sets, and any pipeline that generates a series rather than a single hero image, the consistency across edits is worth more than the last few percent of fidelity.

How to use the templates below

Each card is a prompt that produced the image shown. Open one to read the full prompt, copy it, and swap the parts you care about — subject, palette, setting, camera angle. Prompts that lean on precise text rendering are better served by GPT Image.

Supported settings

Capabilities
Text to imageImage editingMulti-image referenceCharacter consistencyUp to 4K
Resolutions
1k2k4k
Aspect ratios
1:12:33:23:44:39:1616:9

Pitfalls to avoid

  • Nano Banana rewards concrete nouns over adjectives. "A cracked ceramic mug on wet slate" beats "a beautiful artistic mug".
  • It has no quality tier. If output looks soft, raise the resolution or tighten the prompt — there is no knob to turn.
  • Character consistency comes from reference images, not from repeating a name. Feed the same reference on every edit in a series.
  • It follows the last instruction it reads when two instructions conflict, so put the non-negotiable constraint at the end.
  • Text inside images is its weakest area. For posters and packaging with real copy, GPT Image is the safer pick.

Nano Banana prompt templates

Featured Nano Banana prompts

Editable templates are on the way. Until then, these are the cases worth starting from.

Latest Nano Banana cases

Browse all Nano Banana prompts