The Nano Banana 2 Price in 2026: Cost per Image and API

The Nano Banana 2 price depends on one setting more than anything else: the resolution of the image you ask for. On Unicorn API, a Nano Banana 2 image costs 8 credits at 1K, 12 credits at 2K and 18 credits at 4K. This post turns those credits into dollars, compares them with the other Google image models (Nano Banana Pro, Nano Banana 2 Lite and Nano Banana 2.1), works through what a real batch of 1,000 images costs, and shows the API code to generate and price images yourself.

What Is the Nano Banana 2 Price per Image?

Nano Banana 2 is Google’s current Gemini image model, released on 26 February 2026. It handles text-to-image generation and image edits, renders text inside images well, and accepts up to 14 reference images (pictures you pass in so the model can match a product, a face, a style or a layout). It outputs at three resolutions: 1K, 2K and 4K. “1K” means the longer side is roughly 1,000 pixels; 4K is roughly four times that on each side.

Here is the full Nano Banana 2 price list on Unicorn API:

Resolution	Credits per image	Monthly plan ($0.01/credit)	Top-up ($0.015/credit)
1K	8	$0.08	$0.12
2K	12	$0.12	$0.18
4K	18	$0.18	$0.27

Three details matter when you read this table:

• You pay per output image. If you set the input field n to 4, you get four images and pay four times the per-image price. A single 4K request with n: 4 costs 72 credits.
• If you leave resolution unset, you are billed at the 1K rate of 8 credits. Any combination outside the three listed resolutions falls back to 8 credits.
• Aspect ratio does not change the price. A 16:9 banner, a 9:16 story and a 1:1 square at the same resolution all cost the same. Nano Banana 2 supports 16:9, 1:1, 9:16, 21:9, 4:3, 3:2 and 2:3.

Reference images are not part of the price formula either. Passing 14 product shots to guide an edit costs the same as passing none; only the output resolution and the number of outputs count.

How Credits Turn Into Dollars

Unicorn API bills everything in credits, so the same balance pays for image, video, audio and chat models. There are two ways to buy credits, and which one you use changes the effective Nano Banana 2 price by 50%.

Monthly plans: $0.01 per credit

Monthly plans give you credits at $0.01 each. A plan is the cheaper option if you generate images every month.

Plan	Price per month	Credits	1K images	2K images	4K images
Starter	$25	2,500	312	208	138
Creator	$50	5,000	625	416	277
Studio	$100	10,000	1,250	833	555
Pro	$200	20,000	2,500	1,666	1,111

The image counts assume you spend the whole balance on Nano Banana 2 at one resolution. In practice most people mix: drafts at 1K, finals at 2K or 4K. Note that the Pro plan is a billing tier and has nothing to do with the Nano Banana Pro model.

Top-ups: $0.015 per credit, no expiry

Top-up credits cost $0.015 each and never expire. They suit irregular workloads: a one-off campaign, a prototype, or a buffer on top of a plan when a busy month runs over. The trade-off is simple: you pay 1.5x the plan rate in exchange for no commitment. Full details are on the pricing page.

Know the exact cost before you run anything

Every request can be priced in advance with POST /v1/models/price. That is useful when your code builds requests dynamically (for example, a user picks the resolution and the number of variations) and you want to show or cap the cost before spending credits. The request format is in the API documentation.

Nano Banana 2 vs Pro, Lite and 2.1: Price Comparison

Google now has several Nano Banana models, and the names are confusing. Searches like “nano banana 2 pro” and “nano banana pro 2” usually mean one of two separate models: Nano Banana 2 or Nano Banana Pro. There is no model called “Nano Banana 2 Pro”. Here is how the whole family is priced on Unicorn API, per image:

Model	Released	1K	2K	4K	Reference images
Nano Banana	Aug 2025	4 credits (single rate)	n/a	n/a	Up to 3
Nano Banana Pro	Nov 2025	18 credits	18 credits	24 credits	Up to 8
Nano Banana 2	Feb 2026	8 credits	12 credits	18 credits	Up to 14
Nano Banana 2 Lite	Jul 2026	4 credits	n/a	n/a	Up to 10
Nano Banana 2.1	Oct 2026	4 credits	6 credits	9 credits	Via image_input

Nano Banana Pro vs Nano Banana 2

Nano Banana Pro is built for high-fidelity output with accurate text rendering. It costs more than double Nano Banana 2 at 1K (18 credits versus 8), 1.5x at 2K (18 versus 12) and a third more at 4K (24 versus 18). Pro’s 1K and 2K prices are identical, so if you use Pro at all, there is no saving in choosing 1K over 2K.

Nano Banana 2 accepts more reference images (14 versus 8) and two extra aspect ratios (3:2 and 2:3, which Pro lacks). For most product shots, social graphics and edits, Nano Banana 2 is the default choice, and Pro is the model you test when text accuracy or fine detail on a specific job is not good enough.

Nano Banana 2 Lite vs Nano Banana 2

Lite is Google’s cheapest Gemini image model: fast 1K images and edits with up to 10 reference images, at 4 credits per image. That is half the 1K Nano Banana 2 price. Lite has no resolution option, so it is a drafting and volume tool, not a source of print-size finals.

Where Nano Banana 2.1 fits

Nano Banana 2.1 arrived on 7 October 2026 and is priced well below Nano Banana 2 on Unicorn API: 4, 6 and 9 credits at 1K, 2K and 4K, which is half of Nano Banana 2 at every resolution. Its input fields differ: references go in image_input rather than ref_images, and there is no n field, so you request one image per job. If you are setting up a new pipeline today, test 2.1 alongside Nano Banana 2 on your own prompts before committing to either.

Outside the Google family

For context, two non-Google models sit on the same credit balance. GPT Image 1.5 from OpenAI costs 3, 4 or 22 credits at low, medium and high quality, with optional transparent backgrounds. Flux 3 Klein Text to Image from Black Forest Labs costs 1.6, 2.5 or 6.5 credits at 1K, 2K and 4K, with a matching Flux 3 Klein Edit Image model for edits. You can browse everything in the model catalog.

Worked Example: What 1,000 Images Cost

A common question goes like this: “I need 1,000 square images at 2K. What will that cost?” Here is the answer for each candidate model, at both credit rates.

Model (2K, 1:1)	Credits for 1,000	Monthly plan	Top-up
Nano Banana 2	12,000	$120	$180
Nano Banana Pro	18,000	$180	$270
Nano Banana 2.1	6,000	$60	$90
Nano Banana 2 Lite (1K only)	4,000	$40	$60

A Studio plan ($100, 10,000 credits) plus 2,000 top-up credits ($30) covers the Nano Banana 2 job for $130. That is a better deal than a $200 Pro plan if this is a one-off, and cheaper than buying all 12,000 credits as top-ups ($180).

A smarter workflow cuts the cost further. Few teams accept the first result for every prompt, so assume you generate drafts before finals:

1. Draft at 1K. Generate two candidates per prompt at 1K with Nano Banana 2 (16 credits per prompt), or with Lite if speed and cost matter more than detail (8 credits per prompt).
2. Pick a winner. A person or a scoring script picks the better candidate.
3. Re-render the winner at 2K. Run the same prompt and references at 2K for 12 credits.

With Nano Banana 2 drafts, that is 28 credits per finished image, or 28,000 credits ($280 on plan credits) for 1,000 finals. With Lite drafts it drops to 20 credits per image, or $200. Compare that with generating four 2K candidates per prompt and picking one: 48 credits per image, or $480. Iteration, not the headline Nano Banana 2 price, is usually where the money goes.

One caveat: a re-render at a higher resolution is a new generation, not an upscale, so it will not be pixel-identical to the draft. Strong prompts and reference images keep it close. If you need the exact draft enlarged, generate at the final resolution from the start.

How to Generate Nano Banana 2 Images Through the API

All media models on Unicorn API work the same way: you create a job with POST /v1/jobs, poll GET /v1/jobs/{id} until the status is succeeded, then read the image URL from outputs[0].url. Create an API key on the keys page and export it as UNICORN_API_KEY.

curl

curl -X POST https://api.unicornapi.net/v1/jobs \
  -H "Authorization: Bearer $UNICORN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "nano-banana-2",
    "input": {
      "prompt": "A ceramic coffee mug on a walnut desk, morning window light, product photo",
      "resolution": "2K",
      "aspect_ratio": "1:1"
    }
  }'

# Then poll with the id from the response:
curl https://api.unicornapi.net/v1/jobs/JOB_ID \
  -H "Authorization: Bearer $UNICORN_API_KEY"

This request costs 12 credits: one image at 2K.

Python, with polling

import os
import time
import requests

API = "https://api.unicornapi.net"
HEADERS = {"Authorization": f"Bearer {os.environ['UNICORN_API_KEY']}"}


def generate(prompt, resolution="1K", aspect_ratio="1:1", model="nano-banana-2"):
    resp = requests.post(
        f"{API}/v1/jobs",
        headers=HEADERS,
        json={
            "model": model,
            "input": {
                "prompt": prompt,
                "resolution": resolution,
                "aspect_ratio": aspect_ratio,
            },
        },
        timeout=30,
    )
    resp.raise_for_status()
    job = resp.json()

    # Poll until the job finishes (see /docs for every possible status).
    for _ in range(120):
        if job["status"] == "succeeded":
            return job["outputs"][0]["url"]
        if job["status"] == "failed":
            raise RuntimeError(f"Job {job['id']} failed: {job}")
        time.sleep(3)
        job = requests.get(
            f"{API}/v1/jobs/{job['id']}", headers=HEADERS, timeout=30
        ).json()

    raise TimeoutError(f"Job {job['id']} did not finish in time")


# Draft cheaply at 1K (8 credits), then render the final at 2K (12 credits).
prompt = "A ceramic coffee mug on a walnut desk, morning window light, product photo"
draft_url = generate(prompt, resolution="1K")
print("Draft:", draft_url)

final_url = generate(prompt, resolution="2K")
print("Final:", final_url)

To switch models, change the model argument. Nano Banana Pro uses the same input fields (prompt, resolution, aspect_ratio, ref_images, n, output_format), though it only supports 16:9, 1:1, 9:16, 21:9 and 4:3. Lite takes no resolution field, and 2.1 uses image_input for references. To try prompts before writing code, use the image playground; if you work in Cursor or Claude Code, you can call the same models from your editor through the MCP integration.

Mistakes That Inflate Your Image Bill

Most overspending on image generation comes from a handful of habits. Each of these is easy to fix.

• Defaulting to 4K. A 4K Nano Banana 2 image costs 2.25x a 1K image. Web thumbnails, chat avatars and social posts rarely need more than 1K or 2K. Save 4K for print, hero banners and assets you will crop heavily.
• Forgetting that n multiplies. n: 4 at 4K is 72 credits per request. If your app lets users pick a variation count, price the request first.
• Using Pro by habit. Nano Banana Pro is the right call for some text-heavy designs, but at 1K it is 2.25x the Nano Banana 2 price. Test Nano Banana 2 first and move only the jobs that fail to Pro.
• Running steady volume on top-ups. Top-ups cost 1.5x plan credits. If you spend a predictable amount every month, a plan pays for itself immediately; keep top-ups for spikes.
• Comparing prices without resolution. A “per image” price quoted anywhere is meaningless without the resolution and quality level. Always compare 2K with 2K.
• Ignoring rounding on fractional prices. Flux 3 Klein is priced at 1.6 credits per 1K image, and totals are rounded up to whole credits per request. One image costs 2 credits; five images in one request with num_images: 5 cost 8, not 10. Batch where the model allows it. The Nano Banana models are priced in whole credits, so rounding does not affect them.

Which Model Should You Pick?

Price is only half the decision; the other half is whether the output is good enough for your use. A practical rule set:

• High-volume drafts, previews and 1K edits: Nano Banana 2 Lite at 4 credits. Its 10 reference images make it surprisingly capable for consistent product edits.
• General production images and edits up to 4K: Nano Banana 2. It has the widest set of aspect ratios and the most reference images (14) of any Nano Banana model.
• Cost-sensitive work at 2K and 4K: Nano Banana 2.1, at half the Nano Banana 2 price at every resolution. Compare outputs on your own prompts first.
• Text-heavy designs where fidelity matters most: Nano Banana Pro, at 18 credits up to 2K and 24 at 4K.
• Transparent backgrounds: GPT Image 1.5, which supports an optional transparent background.

Because every model draws from the same credit balance and uses the same job API, switching is a one-line change. The cheapest way to settle “Nano Banana 2 vs Pro” for your project is to run 20 of your real prompts through both and look at the results side by side. At 1K that costs 160 credits for Nano Banana 2 and 360 for Pro: $5.20 on plan credits for a decision that will shape every image you generate afterwards.