Skip to main content

Nano Banana 2 image editing takes one to fourteen source images and a description of the change, at 0.5k to 4k. Keep the source shape or choose one of fourteen aspect ratios, with optional web and image search.

From $0.036/image20% below the publisher's list price

Workbench

Input

Image editing
228/5000

Reordering updates prompt references. Removing media also removes its references; review your prompt.

Up to 5,000 characters. Describe subject, setting, light and lens in plain sentences; words in quotes are rendered as text in the image. at least 1 characters · at most 5000 characters
affects price1/14

Drop files or media links, paste, or browse · Up to 14 files · Max 10 MB · JPG · PNG · WEBP · GIF

  • image 1
1 / 14 files
For Nano Banana 2 reframing, send 1–14 sources and describe the intended change before listing the details to preserve. Up to 10 images fit economy processing; 11–14 need Official fallback, official pricing and a separate fee for each input. at least 1 items · at most 14 items
Nano Banana 2 has fourteen aspect ratios reaching 8:1 and 1:8, allowing a draft composition to be explored before choosing a larger output tier; framing itself adds no charge. Nano Banana 2 — For editing, auto keeps the source aspect ratio by default.
affects price
Nano Banana 2 offers 0.5k drafts alongside 1k (default), 2k and 4k; 0.5k requires Official fallback and official pricing.
Official fallback
Official fallback is on initially for Nano Banana 2, enabling official-price execution when economy processing cannot finish. It is also necessary for 0.5K drafts, searches and edits above 10 sources. Turn it off only if you accept economy-only capabilities and possible failed requests.
Safety checkerOff

SpicyAPI does not run a safety checker on this model. Your prompt and media go to the model as written, and the model itself decides what comes back.

Text-only refusals are charged.

Checking your account before enabling submission and pricing.

Output

Ready for a new run
rooftop-redExample result

Missing the model you need?

GOOGLE GEMINI IMAGE MODEL

Nano Banana 2 API at a glance

Google's Nano Banana 2 (Gemini 3.1 Flash Image) with search and extreme banner shapes, starting well below Google's list price.

Initial processing
Economy
Official fallback is on initially for Nano Banana 2, enabling official-price execution when economy processing cannot finish. It is also necessary for 0.5K drafts, searches and edits above 10 sources. Turn it off only if you accept economy-only capabilities and possible failed requests.
Size tiers
0.5K–4K
Four tiers, from quick 512-pixel drafts to 4096-pixel finals.
Widest banner
8:1
Fourteen shapes, including 8:1, 4:1, 1:4 and 1:8.
Reference images
14
Up to fourteen sources in one edit.

Results

What you can make with Nano Banana 2

  • A wide panorama of a couple holding hands and walking barefoot along the surf at dusk under a violet sky

    Extreme shapes

    A 4:1 panorama in one call

    A couple walking barefoot along the surf at dusk, rendered as a 4:1 banner at the 2k tier: 4128 pixels wide, with beach and sky filling both ends.

    Text to image, 2k, 4:1.

  • A woman in an oversized cream blazer over black lace and high-waisted trousers by a hotel window at night

    Fashion

    Editorial looks with texture

    An oversized cream blazer over black lace by a hotel window at night; the lace, the wool and the city lights all read clearly at 1k.

    Text to image, 2k, 3:4.

  • A woman in a red off-shoulder satin dress beside a grand piano at a rooftop pool bar, a turquoise pool and the city skyline at golden hour

    Editing

    New location, same person

    The woman in the red satin dress moved from a jazz lounge to a rooftop pool bar at golden hour, her face, dress and pose kept.

    Edit, one source, 3:4.

Content test

Is Nano Banana 2 uncensored? What we tested

Content levels tested on Nano Banana 2, by mode
LevelImage editingText to image
Mild
RefusedTested Sep 29, 2026
Not tested yet
Moderate
Not tested yet
Not tested yet
Strong
Not tested yet
RefusedTested Sep 29, 2026
Very strong
Not tested yet
Not tested yet
Maximum
Not tested yet
Not tested yet
How we test

How each mode behaves

All modes
Nano Banana 2 has Google content filtering for prompts, source images and outputs, so adult requests can be refused. SpicyAPI tests found refused edits and a generation that returned a billable text-only refusal; other rejections are not charged. SpicyAPI adds no platform filter and cannot unlock Google’s policy. For a different policy, consult the unrestricted Qwen Image Spicy Edit family.
All modes
Results come back as PNG from text to image and as JPEG from editing; the format cannot be chosen.

Use cases

What to build with the Nano Banana 2 API

  • Banners and panoramas

    Website headers, billboards and scroll banners in 4:1 or 8:1, rendered as one continuous scene instead of stretched or stitched.

  • Fast campaign iteration

    Many directions per hour at 0.5k and 1k, then the chosen one again at 2k or 4k.

  • Fashion and lifestyle editorials

    Evening wear, editorial fashion and couple portraits, which rendered in our text-to-image tests.

Prompts

Nano Banana 2 prompt examples you can copy

  • Text to image

    A 4:1 panorama in one call

    An ultra-wide panoramic photograph: a man and a woman walk barefoot along the edge of the surf at dusk, holding hands, she wears a flowing white shirt dress and he wears rolled linen trousers and an open shirt, a long empty beach and a violet sky stretching to both sides.

    Why it works. A 4:1 panorama at 2k. The prompt asks for an ultra-wide photograph and gives both ends of the frame something to hold: a long beach and a wide sky.

    Try this prompt
  • Text to image

    Editorial looks with texture

    Fashion editorial photograph of a woman in an oversized cream blazer worn open over a black lace bralette and high-waisted trousers, leaning on a floor-to-ceiling window of a dark hotel suite at night, city lights behind her, confident look over her shoulder, soft warm lamp light.

    Why it works. A fashion frame at 3:4 and 1k. Wardrobe, window, time of day and lamp light are each named, so the lace and the wool both get detail.

    Try this prompt
  • Image editing

    New location, same person

    Move the woman from image 1 to a rooftop bar at golden hour: the city skyline behind her, a glowing turquoise pool at her side, warm low sun on her face. Keep her face, hair, the red satin dress and her pose exactly as they are.

    Why it works. One source. The prompt names the new setting and light, then lists what stays: face, hair, dress and pose. The 3:4 ratio matches the source framing.

    Try this prompt

Pricing

Nano Banana 2 API pricing

One image starts at $0.036 at 0.5k and $0.09 at 4K.

Get started

How to use the Nano Banana 2 API in three steps

  1. Step 1

    Draft at 0.5k

    The 0.5k tier returns in seconds. It runs only on the official processing option, at the official price, so keep Official fallback on for it; settle the composition there, then run the same prompt at 2k or 4k.

  2. Step 2

    Switch on search for real-world facts

    Web search helps with current details such as weather, dates and landmarks; image search helps with real places and products. Each is off by default and may add a charge per request. Requests with search on run on the official processing option at the official price.

  3. Step 3

    Name each reference’s role

    In edits, say which source supplies the person, which the setting and which the product. With up to fourteen sources the model needs that map.

Parameter reference and code samples are in the API and Readme tabs at the top of this page.

FAQ

Nano Banana 2 API FAQ

About Nano Banana 2

Nano Banana 2 is Google's Gemini 3.1 Flash Image: four size tiers from 0.5K to 4K, fourteen shapes up to 8:1, fourteen reference images and optional web and image search, starting well below Google’s list price. Google’s content policy applies, so it fits fast campaign, fashion and banner work rather than anything Google's policy would block.

Related models and next tasks