Is this Google's Nano Banana 2.1?

Yes. Requests run on the model Google released on 6 October 2026 as Nano Banana 2.1, an update to Nano Banana 2 (Gemini 3.1 Flash Image). The 1K, 2K and 4K outputs match the sizes Google publishes for it.

Nano Banana 2.1 or Nano Banana 2?

2.1 is the newer model: Google improved image quality, prompt adherence, character consistency and text rendering, and says it fixed seams on wide banners at 2K and 4K. Nano Banana 2 keeps a 0.5K draft tier that 2.1 does not have. Both offer web and image search and up to fourteen references. Google has deprecated Nano Banana 2 (no shutdown date announced) and recommends moving to 2.1.

How well does it render text?

Every line we put in quotes came out spelled as written: a magazine masthead with two cover lines, a film title with a release line, a neon motel sign, and an album title designed onto a blank sleeve in an edit. Small print we did not spell out, such as the fine print under a poster title or a barcode, came out as filler.

Can it make NSFW images?

No. Google's content policy applies. Text to image renders the Mild level, but requests from the Strong level up usually come back as a finished image toned down to something milder, and that image is billed like any other, so rephrase rather than resend. Editing refused the Mild-level swap and the stronger edits we tried; refused requests are not charged. The economy route refuses even Mild-level requests, so with Official fallback on those requests run on the official route at the official price.

What do web search and image search do?

Web search lets the model check current facts before drawing; image search lets it look at real pictures of a place or product. Both are off by default, and each adds a charge per request when switched on, shown in the quote. Requests with search on run on the official route at the official price, so they need Official fallback left on.

Why does the quote show two prices?

Every request runs on the economy route first and costs the economy price when it succeeds there. With Official fallback on, the default, a request the economy route refuses or cannot complete runs again on the official route and costs the official price instead, so the quote shows the starting price and the most the request can cost. The higher amount is held when you submit, and the difference is returned if the economy route delivers. Turn Official fallback off to pay only the economy price, accepting that some requests will then be refused or fail.

How big are the images?

At 1:1 the tiers are 1024, 2048 and 4096 pixels square. Other shapes keep about the same pixel count: in our tests 3:4 at 1k came back 896 × 1200, 16:9 at 2k 2752 × 1536 and an 8:1 banner at 2k 5856 × 704.

Nano Banana 2.1 API: Google's Newest Image Model, 1K to 4K

google/nano-banana-2.1/text-to-image

Nano Banana 2.1 text to image is Google's newest Nano Banana model. It renders at 1k, 2k or 4k in fourteen shapes up to 8:1, spells quoted text correctly and can search the web or reference pictures before drawing.

From $0.022/imageWith Official fallback: up to $0.048

Run model View API

Input

Text to image

Prompt559/5000

Aspect ratio

Resolutionaffects price

Web searchaffects price

Enable Image Searchaffects price

Output format

PNGJPEG

Official fallback

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.

Checking your account before enabling submission and pricing.

Output

Ready for a new run

Example result

Animate

Edit

Split into layers

{
  "model": "google/nano-banana-2.1/text-to-image",
  "input": {
    "aspect_ratio": "3:4",
    "resolution": "1k",
    "enable_web_search": false,
    "enable_image_search": false,
    "output_format": "png",
    "official_fallback": true,
    "prompt": "Fashion magazine cover photograph. The masthead across the top, in tall elegant serif capitals, reads \"NOCTURNE\". A woman in her early thirties with a sleek dark bob stands on the curving concrete spiral staircase of a brutalist concert hall at night, wearing a plunging emerald velvet gown with a thigh-high slit, one hand on the brass rail, looking down at the camera. Small cover lines on the left read \"The Night Issue\" and \"Velvet, Concrete & Sound\". A small barcode in the lower right corner. Low-key cinematic lighting, deep green and charcoal palette."
  }
}

Introduction

Nano Banana 2.1 Text to image is Google's newest Nano Banana model, released on 6 October 2026 as an update to Nano Banana 2 (Gemini 3.1 Flash Image). Google improved visual quality, prompt adherence and text rendering, and says it fixed the tiling seams wide banners showed at 2K and 4K. It renders at 1k, 2k or 4k in fourteen shapes up to 8:1 and can search the web or reference pictures before drawing.

Choose a mode by starting material

What you have	Mode
a written brief	Nano Banana 2.1 Text to image
images to change or combine	Nano Banana 2.1 Edit

Key features & highlights

Text that reads correctly

In our tests every line we put in quotes came out spelled as written: a magazine masthead and two cover lines, a film title and its release line, a neon motel sign. Small print we did not spell out, such as the fine lines under a film title, came back as filler.

Fourteen shapes, three sizes

From 1:8 story frames to 8:1 banners, at 1k, 2k or 4k. The shape never changes the price; only the size tier does. Our 8:1 banner at 2k came back 5856 pixels wide. Whether a wide frame comes back as one picture depends on the people in it: landscapes and crowds stayed in one piece in our runs, while people spread across the frame or close to the camera often came back as stitched shots.

Search before drawing

Web search pulls current facts such as dates, weather and events; image search pulls reference pictures of real places and products. Our forecast poster printed today's date with web search on and an invented date without it.

PNG or JPEG

PNG is the default; JPEG files are several times smaller at the same price.

Request parameters

Request parameters for Nano Banana 2.1 · Text to Image
Parameter	Required	Description
`prompt`	Yes	Up to 5,000 characters. Describe subject, setting, light and lens in plain sentences; words in double quotes are printed in the image. at least 1 characters · at most 5000 characters
`aspect_ratio`	No	One of fourteen shapes, 16:9 by default; the shape does not change the price. `1:1` · `3:2` · `2:3` · `3:4` · `4:3` · `4:5` · `5:4` · `9:16` · `16:9` · `21:9` · `1:4` · `4:1` · `1:8` · `8:1` · Default: `16:9`
`resolution`	No	1k (default), 2k or 4k. A square frame is 1024, 2048 or 4096 pixels across; each tier has its own price. `1k` · `2k` · `4k` · Default: `1k`
`enable_web_search`	No	Off by default. When on, the model looks up current facts such as dates, weather or events before drawing. Default: `false`
`enable_image_search`	No	Off by default. When on, the model looks up reference pictures on the web before drawing, useful for real places and products. Default: `false`
`output_format`	No	png (default) or jpeg. JPEG files are several times smaller; the price is the same. `png` · `jpeg` · Default: `png`
`official_fallback`	No	On by default. Each request runs on the economy route first. Default: `true`

Pricing

Billed per image by size tier; the shape and output format do not change the price. Each request starts on the economy route at the economy price. With Official fallback on, the default, a request the economy route refuses or cannot complete runs again on the official route at the official price, so the quote shows both the starting price and the most the request can cost; turn it off to pay only the economy price. Requests with web search or image search run on the official route, and each search adds a charge. A request the model tones down still returns an image and is billed like any other. Refused requests are not charged.

Pricing
Specification	Price
`1k`	$0.0217 / image (With Official fallback: up to $0.048)
`2k`	$0.0351 / image (With Official fallback: up to $0.072)
`4k`	$0.0626 / image (With Official fallback: up to $0.162)

Use cases

Covers and posters

Magazine covers, film posters and event flyers where the headline has to read correctly on the first try.

Website headers

Ultra-wide 4:1 and 8:1 headers of landscapes, harbours, skylines and crowds, checked for stitched panels before use. For people up close, render 21:9 at 4k and crop to 4:1. Tall 1:4 and 1:8 frames work when the subject itself is tall, such as a lighthouse, a skyscraper or a waterfall.

Fashion and lifestyle editorials

Evening wear, editorial fashion and couples render as asked; requests from the Strong level up do not.

Pictures that depend on today

Forecast cards, schedules and event posters, with web search on so the details are current.

Explore key use-case tutorials

Prompt guide & pro tips

Quote the words, describe the type

Put each line of text in double quotes and say its style and place: tall serif capitals across the top, bold condensed sans serif at the bottom.

Keep people small in wide frames

For 4:1 and 8:1, let one setting fill the whole width, such as a ridge, a harbour or a crowd, and keep any people small and near the centre. People spread across the frame or close to the camera were the ones that came back as stitched shots; opening with "one continuous scene" did not prevent it.

Keep search for facts

Web search helps when the picture depends on something current or real. It does not improve style or composition, so leave it off otherwise.

Draft at 1k, deliver at 2k

Settle the composition at 1k, then run the same prompt at 2k; keep 4k for print and heavy crops.

API quick start

# Generate the idempotency key once; reuse it when retrying after a lost response.
export SPICY_IDEMPOTENCY_KEY="${SPICY_IDEMPOTENCY_KEY:-$(uuidgen)}"
curl -X POST https://api.spicyapi.ai/api/v1/jobs/createTask \
  -H "Authorization: Bearer $SPICY_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: $SPICY_IDEMPOTENCY_KEY" \
  -d '{
  "model": "google/nano-banana-2.1/text-to-image",
  "input": {
    "prompt": "A ceramic teapot on a linen cloth, soft window light, muted palette",
    "resolution": "1k",
    "aspect_ratio": "16:9",
    "output_format": "png",
    "enable_web_search": false,
    "enable_image_search": false
  }
}'

# Copy data.taskId from the response into TASK_ID, then poll the status.
TASK_ID="TASK_ID_FROM_RESPONSE"
curl "https://api.spicyapi.ai/api/v1/jobs/recordInfo?taskId=$TASK_ID" \
  -H "Authorization: Bearer $SPICY_API_KEY"

See the API reference for authentication, the complete schema, polling, and error handling.

Missing the model you need?

Tell us what to add and where you would use it. Requests go to the people who decide what gets onboarded next.

GOOGLE IMAGE API

Nano Banana 2.1 API at a glance

Google's newest Nano Banana: words spelled right, faces kept across edits, and search-grounded images from 1K to 4K.

Size tiers: 1K–4K; Three tiers; a square frame is 1024, 2048 or 4096 pixels across.
Widest shape: 8:1; Fourteen shapes, from 1:8 story frames to 8:1 banners.
Reference images: 14; Up to fourteen sources in one edit.
Search options: 2; Web search for current facts, image search for real places and products.

Results

What you can make with Nano Banana 2.1

Text in the image
A cover with every word spelled right
A magazine cover with the masthead NOCTURNE, two cover lines and a barcode, around a woman in an emerald velvet gown on a concrete spiral staircase. Every quoted word came out as written.
Text to image, 2k, 3:4.
Three references
One face, one jacket, one room
A studio headshot, a product shot of a suede overshirt and an empty observatory combined into one night scene, keeping his face and the jacket's brass snaps and chest pockets.
Edit, three sources, 4:5.
Extreme shapes
An 8:1 banner built around one crowd
A packed warehouse rave at 8:1 on the 2k tier, 5856 pixels wide, with a couple dancing forehead to forehead in the middle of the crowd. This run came back as one scene, but people this close to the camera came back stitched in about half of our wide runs; crowds, small distant figures and landscapes are the dependable choice.
Text to image, 2k, 8:1.

More clips made with Nano Banana 2.1

Edit, one source, auto ratio.
Text to image, 2k, 2:3.
Text to image, 2k, 21:9.
Edit, one source, auto ratio.

Content test

Is Nano Banana 2.1 uncensored? What we tested

We ran each level on our live setup and checked every clip frame by frame. A pass means the content was actually on screen, not just that the request went through.

Content levels tested on Nano Banana 2.1, by mode
Level	Image editing	Text to image
MildThe lightest prompts in the test	RefusedTested Oct 7, 2026	Not tested yet
ModerateWhere filtered models start to refuse	Not tested yet	Not tested yet
StrongThe middle of the scale	Not tested yet	SoftenedTested Oct 7, 2026
Very strongWhere most models start to hold back	Not tested yet	Not tested yet
MaximumThe top of the scale	Not tested yet	Not tested yet

Renders: on screen as written. Softened: generated, but milder than what was written. Refused: rejected. Not tested yet: we have not run that level on that mode.

How each mode behaves

All modes: Google's content policy applies to every prompt and result. In our tests text to image rendered the Mild level, but requests from the Strong level up mostly came back as a finished image toned down to something milder, billed like any other; a few were refused outright, as were the Maximum-level requests. Editing refused the Mild-level swap and the stronger edits we tried, returned one bolder request toned down (billed), and let one Maximum-level edit through. Refused requests are not charged. The economy route adds a stricter check of its own and refuses anything from the Mild level up: with Official fallback on, such a request runs again on the official route at the official price; with it off, the request is refused and not charged.
All modes: There is no seed and no negative prompt, so the same request gives a different image each time. The smallest tier is 1K; there is no 0.5K draft size.
All modes: Extreme shapes are not always one picture, and the people in the scene decide it. In our runs, 4:1 and 8:1 banners with people spread across the frame came back as stitched shots every time, and people close to the camera about half the time, while crowds, small distant figures and landscapes came back as one continuous scene. Tall 1:4 and 1:8 frames worked for subjects that are tall themselves, such as a lighthouse on a cliff, a skyscraper or a waterfall, but a mountain ridge squeezed into 1:4 came back stacked or on its side. Check wide and tall results before you use them.

Use cases

What to build with the Nano Banana 2.1 API

Covers, posters and sleeves
Magazine covers, film posters and album art where the title has to read correctly the first time.
Product and outfit composites
Put a garment or product from one photo on a person from another, in a setting from a third, with the details kept.
Wide banners and tall frames
4:1 and 8:1 headers of landscapes, harbours, skylines and crowds, which stayed one continuous scene in our runs, and 1:4 or 1:8 frames of a subject that is tall itself. For people up close in a wide frame, render 21:9 at 4k and crop it to 4:1.
Fashion and lifestyle editorials
Evening wear, editorial fashion and couples, which rendered in our text-to-image tests.

Prompts

Nano Banana 2.1 prompt examples you can copy

Paste one into the playground as it is, then change one thing at a time.

Text to image
A cover with every word spelled right
Fashion magazine cover photograph. The masthead across the top, in tall elegant serif capitals, reads "NOCTURNE". A woman in her early thirties with a sleek dark bob stands on the curving concrete spiral staircase of a brutalist concert hall at night, wearing a plunging emerald velvet gown with a thigh-high slit, one hand on the brass rail, looking down at the camera. Small cover lines on the left read "The Night Issue" and "Velvet, Concrete & Sound". A small barcode in the lower right corner. Low-key cinematic lighting, deep green and charcoal palette.
Why it works. Each piece of text is in quotes with its style and position, and the scene, wardrobe and light are named separately, so the type and the photograph do not compete.
Try this prompt
Text to image
An 8:1 banner built around one crowd
One continuous ultra-wide panoramic photograph, a single unbroken scene from edge to edge, inside a packed warehouse rave at three in the morning: green laser beams cut through the haze above a dancing crowd that fills the whole frame under steel roof trusses. In the middle of the crowd two men in their thirties, a gay couple, dance forehead to forehead, one in an open black mesh shirt and one shirtless with a thin gold chain, sweat glinting under the lasers.
Why it works. One setting, the crowd, fills the whole width, and the couple sits in the middle of it. In our runs the people decided whether a banner stitched: figures spread across the frame or close to the camera often came back as stitched shots, while crowds, small distant figures and landscapes stayed in one piece. Opening with "one continuous" did not change that.
Try this prompt
Image editing
One face, one jacket, one room
Combine the three images: the man from image 1 wears the cognac suede overshirt from image 2, unbuttoned over a white undershirt, and stands beside the brass telescope in the observatory from image 3 at night, one hand on the eyepiece, starlight from the open dome and the warm desk lamp on his face. Keep his face, and keep the jacket's colour, brass snaps and chest pockets exactly.
Why it works. Each source is named by its position and its role, and the details that must survive, the face and the jacket's hardware, are listed at the end.
This mode starts from your own image: add it in the playground before you run.
Try this prompt

Pricing

Nano Banana 2.1 API pricing

One image starts at $0.022 at 1K and $0.063 at 4K.

With Official fallback: up to $0.048–$0.162

The playground shows the exact price for your settings before you run. Failed or timed-out tasks are usually not charged. Some models charge for content refusals as stated on their model page; check those billing terms before submitting.

See every price tier

Get started

How to use the Nano Banana 2.1 API in three steps

Step 1
Put every printed word in quotes
Write the exact text in double quotes and say how it should look and where it sits: a serif masthead across the top, a condensed title at the bottom. In our tests every quoted line came out spelled as written; unquoted small print, such as the fine lines under a poster title, did not.
Step 2
Switch on web search for anything dated
A forecast, a schedule or this week's event needs web search. Our forecast poster printed today's date with it on; the same prompt without search invented a date.
Step 3
Number your references
In a multi-image edit, refer to the first, second and third image and say what each one contributes. Our three-image edit kept the man's face from the first and the jacket's brass snaps from the second.

Open the playground Get an API key

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

Why SpicyAPI

Why run Nano Banana 2.1 on SpicyAPI

No filter from us
We add no content filter of our own. What the model renders is what you get back.
Tested, not guessed
Content labels come from real generations we ran and checked, with the test date on the page.
Pay per result, in dollars
No subscription and no credits. Failed or timed-out tasks are usually not charged. Some models charge for content refusals as stated on their model page; check those billing terms before submitting.
One key for every model
One API key and balance cover image, video, audio and chat models, so you can explore more models from the same account.

FAQ

Nano Banana 2.1 API FAQ

About Nano Banana 2.1

Nano Banana 2.1 is Google's newest Nano Banana: quoted text spelled right, faces and products kept across edits, optional web and image search, fourteen shapes up to 8:1 and three sizes up to 4K. Google's content policy applies, so it suits covers, posters, product composites and fashion work rather than anything past Google's content line.

All Nano Banana models3

Nano Banana 2.1CurrentText to image · Image editingGoogle official price: $0.034From $0.022per imageWith Official fallback: up to $0.048
Nano Banana 2Text to image · Image editingGoogle official price: $0.045From $0.036per image
Nano Banana ProText to image · Image editingGoogle official price: $0.134From $0.067per imageWith Official fallback: up to $0.107

Related models and next tasks

Related comparisons

Nano Banana 2.1 vs Nano Banana 2

Workbench

Input

Output

Introduction

Choose a mode by starting material

Key features & highlights

Text that reads correctly

Fourteen shapes, three sizes

Search before drawing

PNG or JPEG

Request parameters

Pricing

Use cases

Covers and posters

Website headers

Fashion and lifestyle editorials

Pictures that depend on today

Prompt guide & pro tips

Quote the words, describe the type

Keep people small in wide frames

Keep search for facts

Draft at 1k, deliver at 2k

API quick start

A cover with every word spelled right

One face, one jacket, one room

An 8:1 banner built around one crowd

More clips made with Nano Banana 2.1

How each mode behaves

Covers, posters and sleeves

Product and outfit composites

Wide banners and tall frames

Fashion and lifestyle editorials

A cover with every word spelled right

An 8:1 banner built around one crowd

One face, one jacket, one room

Put every printed word in quotes

Switch on web search for anything dated

Number your references

No filter from us

Tested, not guessed

Pay per result, in dollars

One key for every model

Is this Google's Nano Banana 2.1?+

Nano Banana 2.1 or Nano Banana 2?+

How well does it render text?+

Can it make NSFW images?+

What do web search and image search do?+

Why does the quote show two prices?+

How big are the images?+

About Nano Banana 2.1

All Nano Banana models3

Related models and next tasks

Related comparisons