AI image model

Nano Banana 2 AI Image Generator

Choose Nano Banana 2 when the brief, source images, output tier, and framing all need to be selected explicitly before you generate.

Nano Banana 2 task directions

These local image examples are attributed to the Nano Banana family. They show task directions, not a benchmark between model versions.

Every card identifies whether it is a verified model-family example or a visual reference. Reference cards are not model benchmarks or before-and-after claims.

Input contract
Text prompt + up to 5 reference images
Current controls
Prompt length: Up to 20,000 characters · Output resolution: 1K / 2K / 4K · Framing: 15 aspect-ratio choices
Selection boundary
Compare Nano Banana 2 Lite when provider-selected output is acceptable. Compare Nano Banana Pro when more than the standard reference-image allowance is important, or GPT Image 2 when the current 1K framing menu is the main constraint.
Local visual reference for subject and setting.

Subject and setting

Use one clear subject and setting to test whether the written brief establishes the right visual starting point.

Task direction
Use one clear subject and setting to test whether the written brief establishes the right visual starting point.

Verified model-family example

Verified Nano Banana family example from the local ImageStyle prompt library.

Create with Nano Banana 2
Local visual reference for composition direction.

Composition direction

Use a deliberate frame and hierarchy to decide what the prompt or reference must preserve.

Task direction
Use a deliberate frame and hierarchy to decide what the prompt or reference must preserve.

Verified model-family example

Verified Nano Banana family example from the local ImageStyle prompt library.

Create with Nano Banana 2
Local visual reference for detail and material cue.

Detail and material cue

Specify the visible detail that should guide the result instead of treating the reference as a performance test.

Task direction
Specify the visible detail that should guide the result instead of treating the reference as a performance test.

Verified model-family example

Verified Nano Banana family example from the local ImageStyle prompt library.

Create with Nano Banana 2
Create
AI images
Input
Text prompt + up to 5 reference images
Output
1K / 2K / 4K

Compare available settings with related models

Check whether the input, output, and framing controls fit the job before you open a workspace. This table does not treat the example media as a cross-model performance test.

What you can control with Nano Banana 2

Plan the output before generating

Nano Banana 2 exposes 1K, 2K, and 4K selections alongside 15 current framing choices, so the intended delivery format can be decided up front.

Add reference images when they help

Nano Banana 2 currently supports up to 5 reference images. Only upload source material you are allowed to use.

Choose the available output settings

Current output tiers are 1K / 2K / 4K, with 15 aspect-ratio choices. Use the options shown in the workspace before generation.

Current Nano Banana 2 settings

These settings come from the current ImageStyle model configuration and help you decide whether the workflow fits the job.

SettingCurrent availability
InputText prompt + up to 5 reference images
Prompt lengthUp to 20,000 characters
Output resolution1K / 2K / 4K
Framing15 aspect-ratio choices

AI image model

When to choose Nano Banana 2

Reference-guided images with manual 1K, 2K, or 4K selection and broad framing.

Current workflow fit

Choose Nano Banana 2 when the brief, source images, output tier, and framing all need to be selected explicitly before you generate.

When to compare another model

Compare Nano Banana 2 Lite when provider-selected output is acceptable. Compare Nano Banana Pro when more than the standard reference-image allowance is important, or GPT Image 2 when the current 1K framing menu is the main constraint.

How to create an image with Nano Banana 2

Set the output tier and frame as part of the creative brief, then add references only for details that need a visible starting point.

  1. 1

    Choose the delivery frame first

    Select the output tier and aspect ratio that match the placement, then write the prompt around that composition.

  2. 2

    Add references when needed

    Upload approved reference images only when a defined visual starting point will help the result.

  3. 3

    Choose output controls

    Choose the result that fits the job from the current 1K / 2K / 4K options and 15 aspect-ratio choices.

  4. 4

    Generate and review

    Before public use, inspect visible text, product details, people, marks, and framing in the result.

Checks before you create and publish

Treat the workspace as current
Model availability and settings can change. Use the options shown in the current workspace before you submit.
Use approved source material
When you upload a reference image, confirm that you are allowed to use and publish it for the current project.
Review the generated result
Manually check any visible text, product information, and brand element that must be accurate before publishing.

Nano Banana 2 FAQ

What should I know about Nano Banana 2 vs Nano Banana 2 Lite?
Nano Banana 2 lets you select 1K, 2K, or 4K before generation. Nano Banana 2 Lite accepts prompts and references too, but its output tier is selected by the provider.
How is Nano Banana 2 different from Nano Banana 2 Lite?
Nano Banana 2 currently lets you select 1K, 2K, or 4K before generation. Nano Banana 2 Lite uses provider-selected output instead.
Can I use reference images with Nano Banana 2?
Yes. The current workflow supports up to 5 reference images. Only upload images you are allowed to use.
What output settings does Nano Banana 2 currently offer?
Current output tiers are 1K / 2K / 4K, with 15 aspect-ratio choices. The generation workspace is the source of truth for the options available at submission.
What should I prepare before using Nano Banana 2?
Prepare a clear visual direction. If the current model supports reference images, add approved images only when they help preserve a visual starting point.
Can I publish a generated result without review?
No. Before public use, manually review text, product details, people, marks, and any other element that must be accurate.

Updated:

Capability source: Current ImageStyle image-model configuration.

Create with Nano Banana 2