Google — Gemini 2.5 Flash Image (Nano Banana)

Nano Banana

Name: Nano Banana
Brand: google
Availability: InStock

Fast still generation and edits from natural language — built for quick creative exploration.

Best for rapid concepting, lightweight campaign visuals, and clean reference boards before video.

Text→ImageImage→Image (Edit)Fast tierMulti-image blendsCharacter consistencyLow-latency drafts

Generate an image Open the image lab

Pay-as-you-go · Price shown before you generate

Nano Banana AI video example: Nano Banana demo still from MaxVideoAI — Price$0.05/image
Duration8s max
Format9:16 / 16:9 / 1:1 / 4:5 / 5:4 / 4:3 / 3:4 / 3:2 / 2:3 / 21:9

Best use cases

Rapid concepting & mood explorationQuick edits (restyle / relight / swap details)Simple composites & multi-image blendsThumbnails, headlines, and basic layout textProduct mockups & quick SKU variantsReference boards / keyframes before video

Why Nano Banana is powerful

Speed-first workflow (Built to iterate fast when you need lots of options quickly.)
Edits without heavy tooling (Describe the change in plain English; keep the rest consistent.)
Multi-image creativity (Blend references to stay closer to your intent across variants.)
Great “draft layer” for production (Generate boards, then hand off winners to high-end video engines.)

Technical overview

The limits that shape your renders.

Price / image$0.05/image

Text-to-Image

Image-to-Image

Resolution optionssquare_hd / landscape_hd / portrait_hd

Aspect ratios9:16 / 16:9 / 1:1 / 4:5 / 5:4 / 4:3 / 3:4 / 3:2 / 2:3 / 21:9

Output formatjpg / jpeg / png / webp

Release dateAug 2025

Technical overview

Input types: Text prompts, reference images (JPG/PNG/WebP), and edit mode for restyles or cleanups.
Outputs: High-res still images tuned for photorealism and text legibility.
Modes: Text-to-image; image-to-image for restyle, cleanup, and background replacement.
Typical use: Prepare 1–4 keyframes before animating in Sora, Veo, Kling, or Wan.

How to Write a Great Nano Banana Prompt

Nano Banana (Gemini 2.5 Flash Image) rewards short, literal briefs. Lead with composition, then lighting/style. For edits, say what to change and what must stay the same. For multi-image blends, assign each reference a role.

Tip: resolution + aspect ratio are set in the UI — your prompt controls subject, composition, lighting, style, and edit instructions.

Quick prompt (fast iteration)

Use 1–2 sentences when you want variations.

Quick = variations. Use for fast iteration.

Template (copy/paste)

[Subject] in [setting]. [Composition]. [Lighting]. [Style anchor].

If editing:
Edit: change [X]. Keep [Y] (framing, lighting, identity).

If using multiple images:
Refs: Image1 = [main subject]. Image2 = [style/lighting]. (Optional) Image3 = [background/prop].

Structured prompt (best for reliable results)

Separate information so the model can follow it consistently.

Structured = consistency. Use when you need reliable results.

Template (copy/paste)

Subject:
[Who/what + 2–3 defining details.]

Composition:
[Framing + angle + background + what stays clean.]

Lighting:
[Key light + mood + 2 palette anchors.]

Style:
[Photoreal / illustration + texture/grain + one style reference.]

Text (optional):
[Exact copy in quotes + placement + contrast.]
Example: “SALE — 20% OFF” top-left, one line, high contrast.

Edit (if editing):
Change:
- [list the changes]
Keep:
- framing, lighting, identity, logo-free, clean background

References (if using multiple images):
Image1 controls subject/identity
Image2 controls lighting
Image3 controls palette/background

Pro prompt (ultra-specific "studio brief")

Use this when you need a very specific look or repeatable results across a set.

Pro = repeatability. Use for controlled batches.

Template (copy/paste)

Brief:
[Goal + output use case.]

Subject + materials:
[Details + textures + constraints.]

Layout:
[Grid + spacing + focal point + negative space.]

Lighting + grade:
[Key light + contrast + palette anchors.]

Typography (optional):
[Font vibe + hierarchy + exact copy + placement.]
Keep copy short. Avoid paragraphs.

Edits (if needed):
Change list:
- [A]
- [B]
Keep list:
- framing, lighting, identity, background, color palette

Tips & limits

Image generation is most predictable when you keep the frame simple, readable, and physically grounded.

What works best

Fast photoreal stills for moodboards, thumbnails, and keyframes
Edit mode for cleanup, relight, and background swaps (no masking)
Multi-image blends when you assign roles to references
Short, readable text (headlines > paragraphs)

Hard limits to keep in mind

Image-only (no motion)
Edits require an input image (upload at least one reference)
Small on-screen text is hit-or-miss — overlay critical copy in post

Nano Banana vs Nano Banana Pro

View Nano Banana Pro details →

Use Nano Banana when you want:

Fast drafts and quick edits
Rapid concepting and exploration
Lightweight layout tests

Use Nano Banana Pro when you need:

Clean typography and layouts
Consistent product families
High-res finals for campaigns

Safety & people / likeness

No sexual content, and nothing involving minors.
No hateful, harassing, or graphic-violence content.
Don’t impersonate real people or public figures; use consent for any likeness.
Don’t include private personal data (faces of non-consenting people, addresses, phone numbers, documents).
Some prompts or reference images may be blocked by provider safety filters.

FAQ

Is Nano Banana for images or video?

Images. Use it for photoreal references, thumbnails, and keyframes before sending shots to video engines.

How many images per run?

Generate multiple outputs per run (controlled by "num images" / num_images). Keep batches small for faster iteration.

Does edit mode need masks?

No. Upload one or more images and describe the edit in plain language (what to change, what to keep).