Nano Banana 2.1 API Pricing: Cost per Image at 1K, 2K and 4K

Nano Banana 2.1 API pricing on Google's Gemini API starts at $0.0336 for a 1K image and rises to $0.0756 at 4K, because Google bills the image as output tokens and a bigger image uses more of them^[1]. That is roughly half of what Nano Banana 2 cost at the same sizes, which matters now that Google has deprecated Nano Banana 2 and scheduled its Gemini API shutdown for October 29, 2026^[2].

This guide breaks down how the per-image cost is calculated, where the numbers in Google's own documentation disagree, what grounding and thinking add, and how a flat per-image price changes which resolution you should pick. Prices below are Google's list prices as published on October 7, 2026.

TL;DR

Google lists Nano Banana 2.1 at $0.0336 (1K), $0.0504 (2K) and $0.0756 (4K) per image on the Standard tier, and half that on Batch^[1].
The image is billed as output tokens at $30 per million: 1,120 tokens at 1K, 1,680 at 2K, 2,520 at 4K^[1]. Google's Vertex documentation lists 4K at 3,780 tokens, which would make a 4K image about $0.113^[3].
At 1K, Nano Banana 2.1 costs about 50% of Nano Banana 2's $0.067 list price^[1].
Google's free tier is not available for this model^[1].
On reAPI, Nano Banana 2.1 is one flat price per image at 1K, 2K and 4K, with references, grounding and thinking levels included. The live rate is on the model page^[4].

How Google prices Nano Banana 2.1

Google does not sell Nano Banana 2.1 images at a fixed price. It meters three kinds of tokens and adds them up^[1]:

Item	Standard	Batch
Input (text, image, video)	$1.50 / 1M tokens	$0.75 / 1M tokens
Text and thinking output	$7.50 / 1M tokens	$3.75 / 1M tokens
Image output	$30 / 1M tokens	$15 / 1M tokens

The image output line is the one that dominates. Google converts it to per-image figures on the same page:

Resolution	Output tokens	Standard per image	Batch per image
1K	1,120	$0.0336	$0.0168
2K	1,680	$0.0504	$0.0252
4K	2,520	$0.0756	$0.0378

Those per-image numbers are the image tokens only. Your prompt and any reference images are input tokens on top, and the model's reasoning is billed as thinking output^[1]. Vertex AI documents a reference image as 1,120 input tokens^[3], so at $1.50 per million each reference adds about $0.0017. For a single 1K image with no references, input is a rounding error; for a 14-reference fusion job it adds up to about $0.024 before any thinking tokens.

The 4K figure is not settled

Google's two documentation sites disagree on 4K. The Gemini API pricing page uses 2,520 output tokens for a 4K image ($0.0756)^[1]. The Vertex AI model page lists 3,780 tokens^[3], which works out to 3,780 × $30 / 1,000,000 = $0.1134. If you plan a 4K-heavy workload on Google directly, check your first invoice rather than budgeting from either page.

Grounding and thinking

Google Search grounding is billed separately: Gemini 3.x models share 5,000 free search requests per month, then $14 per 1,000^[1]. Google charges for each search query, and one request can trigger more than one^[1]. So after the free allowance a grounded image adds at least $0.014, which is 42% or more on top of a 1K image.

Thinking is billed as output text at $7.50 per million tokens^[1]. Nano Banana 2.1 has three thinking levels (minimal, medium, high) and defaults to medium^[5]. In three of our own requests on October 7, 2026, thinking used no tokens at minimal, 892 at the default level and 2,658 at high. The prompts differed, so treat these as a rough range rather than a benchmark. At $7.50 per million, 2,658 tokens is about $0.02, more than half the price of the 1K image itself.

Nano Banana 2.1 vs Nano Banana 2 and Pro on price

Google priced the update well below its predecessor. The image output rate dropped from $60 to $30 per million tokens^[1]:

Model	1K	2K	4K
Nano Banana 2.1	$0.0336	$0.0504	$0.0756
Nano Banana 2	$0.067	$0.101	$0.151
Nano Banana Pro	$0.134	$0.134	$0.24

All figures are Google Standard-tier list prices per image^[1]. At 1K, Nano Banana 2.1 is $0.0336 / $0.067 ≈ 50% of Nano Banana 2 and 25% of Nano Banana Pro.

Price is only half of the switch decision. Google's changelog marks Nano Banana 2 (gemini-3.1-flash-image) as deprecated with a shutdown date of October 29, 2026, and tells developers to migrate to Nano Banana 2.1^[2]. If you call Google directly, the cheaper model is also the one that will keep working. For a capability comparison, see our Nano Banana 2.1 vs Nano Banana 2 breakdown.

What changes the bill on reAPI

On reAPI the model id is nano-banana-2.1 and the bill has one dimension: the number of images. Each request returns one image and costs one flat price^[4]:

credits = ceil(per_image_price × 1000)     1 credit = $0.001

These do not change the price on reAPI:

Resolution: 1K, 2K and 4K cost the same.
Reference images: zero or fourteen, same price.
Google Search and image search grounding.
Thinking level, from minimal to high.

Failed generations are refunded automatically, credits never expire, and there is no monthly subscription. The current per-image rate is on the Nano Banana 2.1 model page, which reads it from the live price table.

The practical effect is that a flat rate rewards the expensive settings. On Google's meter, a 4K image with high thinking and grounding costs several times a minimal 1K image. On a flat rate it costs the same, so the resolution you choose comes down to how the image will be used rather than to cost.

Getting a Nano Banana 2.1 API key and a first request

The SERP for this model is full of "api key", "api example" and "api tutorial" searches, so here is the shortest working path on reAPI:

Create an account and open API keys in the dashboard. New accounts receive 100 credits ($0.10) for testing.
Send a request:

curl https://reapi.ai/api/v1/images/generations \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "nano-banana-2.1",
    "prompt": "A clean infographic poster titled \"HOW COFFEE IS MADE\" with four labeled steps",
    "aspect_ratio": "3:4",
    "resolution": "2k"
  }'

The response carries a task id. Poll GET /api/v1/tasks/{id} until status is completed, then read output.image_urls. In our tests an image was ready in about 35–45 seconds.

The same request takes image_urls (up to 14 public URLs) for editing, google_search and google_image_search for grounding, and thinking_level for reasoning depth. The full parameter list is in the API docs^[5].

FAQ

What is the API pricing for the Gemini Nano Banana 2 image?

For Nano Banana 2.1, Google's Standard tier lists $0.0336 per 1K image, $0.0504 at 2K and $0.0756 at 4K. The previous Nano Banana 2 was $0.067 at 1K and $0.151 at 4K^[1]. On reAPI, Nano Banana 2.1 is a single flat rate shown on the model page.

Can I use nano banana 2 for free?

Not through Google's paid API: the free tier is listed as not available for Nano Banana 2.1^[1]. On reAPI, new accounts get 100 credits ($0.10), enough to test Nano Banana 2.1 before topping up. That is a test allowance, not a free plan.

Where can I find the API key for Gemini Nano Banana 2?

On Google, keys are created in Google AI Studio. On reAPI, open the API keys page in the dashboard and create one; the same key works for Nano Banana 2.1 and every other model on the platform.

Nano banana 2.1 api free

There is no free Nano Banana 2.1 API from Google^[1]. Batch mode halves the per-image price if you can wait for results. On reAPI the cost is a flat per-image rate with automatic refunds for failed generations.

Nano banana 2.1 api example

The curl request above is a complete example: model, prompt, aspect_ratio and resolution, then poll the task. The API docs have the same request in Python, Node.js and Go.

nano banana 2.1 api pricing in india

Google publishes Gemini API prices in US dollars^[1], and reAPI credits are also priced in dollars ($1 = 1,000 credits). The per-image rate does not change by country.

Choosing a resolution when the price is flat

Nano Banana 2.1 API pricing on Google's meter pushes you toward 1K and minimal thinking, because every step up costs more. That is a sensible default when you pay per token and generate in bulk.

When the price is flat, the calculation flips. Pick 2K or 4K whenever the image will be printed, cropped or used as a hero banner, turn on thinking_level: "high" for infographics and dense text, and use grounding when the picture depends on current facts. The model page shows the current flat rate, and the API docs cover every parameter of the Nano Banana 2.1 API.