How Much Is Nano Banana Pro? API, Plan, and Usable-Image Cost
Nano Banana Pro is $0.134 per 1K/2K image and $0.24 at 4K on Google's API; Google AI Pro is $19.99 a month with no API. LaoZhang API: $0.09 a call at any size.
On this page

As of October 29, 2026, Nano Banana Pro (gemini-3-pro-image) costs $0.134 per 1K or 2K image and $0.24 per 4K image on Google's Standard API rate. Input, thinking and search grounding are billed on top. In the Gemini app, Pro is not sold on its own: it comes as "Redo with Pro" inside a Google AI plan, and the US Google AI Pro plan is $19.99 a month with no API access included.
To call Pro from your own code for less than Google's Standard rate, use a LaoZhang API key with the same gemini-3-pro-image model. It comes from Google's official channel and costs $0.09 per call at 1K, 2K and 4K, for generation and editing alike, which is about 62% below Google's $0.24 at 4K and about 33% below $0.134 at 1K/2K. A single key has no per-key RPM, IPM or concurrency cap. If the image does not need Pro's exact text, Nano Banana 2.1 costs $0.045 per call on the same key. Current prices are on the LaoZhang API models and pricing page.
Disclosure: laozhang.ai is our own service. Its prices are platform pricing, not an independent comparison.
Three things changed since August 2026. Google released Nano Banana 2.1 on October 6, 2026 at about a quarter of Pro's 1K list price, so the real question is now whether a job needs Pro at all. The old gemini-3-pro-image-preview ID was shut down on the Gemini API on June 25, 2026. And the "Flex" row for Pro appears on Google's pricing page but not in Google's Flex documentation, so treat it as unconfirmed for image models.
Three prices, three products
| What you are buying | Billing owner | Billing unit | Where the final answer lives |
|---|---|---|---|
| Manual use in the Gemini app and other consumer surfaces | Google One / Google AI plan | Monthly or annual bundle | Your signed-in checkout |
| Programmatic image generation | Gemini Developer API or Vertex AI project | Image output, inputs, thinking and tools | Official pricing and project bill |
| A compatible or aggregated API | Independent provider | Calls, credits or a provider plan | Provider terms, console and logs |
A Google AI plan does not add API quota or credit to a Developer API project. A provider credit is not a Google token. A plan with "Nano Banana" in its name can be a seller-owned product rather than a Google subscription. Compare prices only after the billing owner and the unit match. For the Lite, 2.1, and Pro API rows side by side with Batch prices and worked budgets, see Nano Banana API Cost: Lite, Nano Banana 2, Pro, Free Reality, and Which Route To Use.
The official API rate card: $0.134 per 1K/2K image, $0.24 at 4K
Google's Gemini API pricing page lists Nano Banana Pro as Gemini 3 Pro Image, model ID gemini-3-pro-image, GA since May 28, 2026 with no shutdown date announced. The preview ID gemini-3-pro-image-preview was shut down on June 25, 2026, according to the Gemini API deprecations page, so code or a provider alias that still sends it will fail rather than bill.
| Gemini API tier | 1K / 2K image output | 4K image output | Text input / thinking (per 1M tokens) | Use it for |
|---|---|---|---|---|
| Standard | $0.134 | $0.24 | $2.00 / $12.00 | Interactive work |
| Batch | $0.067 | $0.12 | $1.00 / $6.00 | Jobs that can wait in a queue |
| Flex (listed, see below) | $0.067 | $0.12 | $1.00 / $6.00 | Not documented for image models |
| Priority | about $0.242 | about $0.432 | $3.60 / $21.60 | Higher-priority execution |
The image prices come from Google's token rule: output at 1K and 2K counts as 1,120 tokens and 4K as 2,000 tokens, at $120 per million image tokens on Standard. Priority has no per-image row, so its figures are calculations from the $216 per million image rate: $216 × 1,120 / 1,000,000 = $0.24192 and $216 × 2,000 / 1,000,000 = $0.432.

The Pro API has no free tier: the free-tier column for every Nano Banana model on the pricing page reads "Not available". Each input image counts as 560 tokens, about $0.0011 on Standard. Pro always thinks before it renders, and those thinking tokens are billed as text output. For Standard and Priority, Google Search grounding includes 5,000 free searches per month shared across Gemini 3.x models, then costs $14 per 1,000 searches, and one request can trigger more than one search.
To add those parts up for your own job, the Nano Banana Pro cost per image calculator takes the size, the tier (Standard, Batch or Priority), thinking tokens and reference images, and shows the cost per image and per 1,000 images with the arithmetic. It does not include search grounding.
Vertex AI (now Gemini Enterprise Agent Platform) charges the same for Pro on the global endpoint: $120 per million image tokens on Standard, $216 on Priority and $60 on Flex/Batch/Off-peak, which works out to the same $0.134, $0.24, $0.067 and $0.12 per image. The Vertex AI pricing page also lists cached input at $0.20 per million tokens, and Vertex lists Pro's retirement as "May 28, 2027 or later".
For request payloads rather than cost, use the Nano Banana Pro API guide. For a price page devoted to the model ID, see Gemini 3 Pro Image API Pricing.
Is Flex half price for Nano Banana Pro? Plan on Batch instead
Google's pages disagree. The Gemini API pricing page shows a Flex row for gemini-3-pro-image at the Batch price, but the Gemini API Flex inference page, last updated September 23, 2026, lists Gemini models such as 3.5 Flash and 2.5 Pro as supported and no image model at all. On Vertex AI, Pro sits in a combined "Flex/Batch/Off-peak" price table, while Nano Banana 2.1 carries an explicit "does not support Flex" footnote.
Batch is the half-price route that Google documents for image models on both platforms. If you try Flex with Pro, check the first project bill before you route volume through it. Queue-friendly workloads are covered in the Nano Banana Pro Batch API cost guide.
Nano Banana Pro vs 2.1 price: 4x at 1K, about 2x at 4K
Nano Banana 2.1 (gemini-nano-banana-2.1) is the update to Nano Banana 2 and became GA on October 6, 2026. Its list price is far below Pro at every size, and Google ranks it above Pro in its own evaluations. Pro still has one clear job.
| Per image, Google list price | 1K | 2K | 4K | Pro costs |
|---|---|---|---|---|
| Nano Banana Pro, Standard | $0.134 | $0.134 | $0.24 | — |
| Nano Banana 2.1, Standard | $0.0336 | $0.0504 | $0.113 | 4.0x / 2.7x / 2.1x as much |
| Nano Banana Pro, Batch | $0.067 | $0.067 | $0.12 | — |
| Nano Banana 2.1, Batch | $0.0168 | $0.0252 | $0.0567 | same ratios |
The ratios are list prices divided: 0.134 / 0.0336 = 3.99, 0.134 / 0.0504 = 2.66 and 0.24 / 0.113 = 2.12. Thinking narrows the gap. Nano Banana 2.1 bills thinking at $7.50 per million tokens and thinks harder by default. In the 9-prompt test of 2.1, Pro and GPT Image 2.5 run on October 8, 2026 at 1K, 2.1 reported 1,209–1,704 thinking tokens on the text-heavy prompts, about $0.009–$0.013 per image at list price. Pro reported 122–252 thinking tokens, about $0.0015–$0.003 at $12 per million. A 1K text job then lands near $0.046 on 2.1 and $0.137 on Pro, about 3x apart rather than 4x.
The quality signals point both ways:
- Exact wording: in that same test, one draw per prompt, Pro kept the requested words and prices on a poster, a Japanese shop sign and a Chinese menu. Nano Banana 2.1 changed something in each of the three: a restyled subline, a stray word on the sign, and a price written out as 三十八元 instead of 38元. All three models drew 单 in its Japanese form 単.
- Overall preference: Google's own model card scores 2.1 above Pro on overall preference, editing and multi-character consistency. On Arena's text-to-image board as of October 6, 2026, 2.1 ranked #5 (preliminary) and Pro at 2K ranked #16. Neither measures whether a specific line of text came out letter-perfect.
Pro is still worth $0.134 when the image carries text that must be exact, such as a sign, label, menu or packaging line, and when a wrong draw costs you a proofread and a reroll. For drafts, product photos, portraits, infographics and simple edits, start with 2.1. Full 2.1 pricing is in Nano Banana 2.1 API Pricing, and model-by-model guidance is in Nano Banana 2 Lite vs 2 vs 2.1 vs Pro.
Turn a list price into a usable-image cost
Generated images are not the denominator a production team cares about. Use:
usable-image cost = total bill / outputs that pass the frozen acceptance card
Take 1,000 attempts at 1K and an illustrative 80% pass rate, so 800 outputs pass. This is a planning example, not a quality claim about any model or provider.
| Option, 1,000 attempts at 1K | Charge before other extras | Accepted outputs | Cost per accepted output |
|---|---|---|---|
| Pro, Google Standard | $134 | 800 | $0.1675 |
| Pro, provider at $0.09 per billed call | $90 | 800 | $0.1125 |
| Nano Banana 2.1, Google Standard | $33.60 | 800 | $0.042 |
On image price alone, 2.1 stays cheaper per accepted 1K image until its pass rate falls below about a quarter of Pro's (0.0336 / 0.134 = 0.25). The break-even is about 38% of Pro's pass rate at 2K and 47% at 4K. That rarely happens on photos. It can happen on exact-text work once review time is counted.
Here is that case with illustrative numbers. Pro passes 800 of 1,000 text-heavy images and 2.1 passes 400. Rejecting a draw takes 30 seconds of a reviewer paid $60 an hour, so $0.50 per rejection. Pro costs $134 + 200 × $0.50 = $234, or $0.2925 per accepted image. Nano Banana 2.1 costs $33.60 + 600 × $0.50 = $333.60, or $0.834 per accepted image. With equal 80% pass rates, 2.1 would cost $33.60 + $100 = $133.60, or $0.167 per accepted image, and win.
Now add input, thinking, grounding, retries, provider charges for failed calls, and human repair. If zero outputs pass, there is no valid denominator. Reporting a zero-dollar usable image in that situation hides a failed setup.

Run five outputs before you commit a budget
Freeze five representative prompts, the required resolution and an acceptance card. If 2.1 is a candidate, run the same five on it. For every result, record:
- requested and returned model ID;
- requested and returned resolution;
usageMetadata(image, thinking and input tokens) or the provider call log;- the actual charge and any grounding or retry activity;
- whether the image passes the acceptance card.
Use a project-defined investigation threshold. A practical starting point is 10% unexplained variance: after documented input, thinking, grounding and retry charges are added, pause the queue if observed cost per attempt stays more than 10% above the estimate. Ten percent is an editorial stop rule, not a Google guarantee.
Two billing controls matter before scale. Gemini API projects on Tier 1 have a spend-based limit of $10 per rolling 10 minutes, which is at most about 74 Standard Pro images at 1K/2K ($10 / $0.134 = 74.6) before a 429 error; Tier 2 and Tier 3 both raise it to $200 (read November 11, 2026; Tier 2 was $50 on October 5), per the Gemini API rate limits page. Prepay credits start at $5, expire after 12 months, and a depleted balance returns a 402 error, according to the Gemini API billing page. Tier requirements and the increase form are in Nano Banana Pro API Quota Increase.
What the $19.99 Google AI Pro card includes
As of October 8, 2026, Google's US Google AI plans page lists Google AI Pro at $19.99 per month with 5 TB of storage and Gemini usage limits four times higher than free. Google AI Ultra starts at $99.99 a month for 5x the Pro limits, or $199.99 for 20x. Google AI Plus sits below Pro at 2x the free limits. None of these plans is a Nano Banana Pro subscription, and none adds API quota or credit. As of October 29, 2026, Google's Google AI plans page still shows the same US prices and no API access in any plan, so if you need Pro inside your own app or script, the subscription does not help and the per-image API prices above apply.
What the plan gives you for images, according to Google's own pages:
- Google's Gemini image generation overview says free users get Nano Banana 2, while Google AI Pro, Plus and Ultra users get Nano Banana 2.1 and can switch a result to Pro with Redo with Pro from the three-dot menu.
- The Gemini Apps image help page still describes Nano Banana 2 as the default image model. Nano Banana Pro there is a paid-plan redo when the Gemini model is set to Pro, and if you reach your daily quota of Nano Banana 2 images, you can't redo more images with Pro. The help pages had not caught up with 2.1 as of October 8, 2026.
- Downloads are 2K with a Google AI plan and 1K without one.
Consumer limits are not a fixed image allowance. The Gemini Apps limits page describes compute-based limits that refresh every five hours until a weekly limit is reached. AI Plus is 2x the standard limit, AI Pro 4x, and AI Ultra 5x or 20x the Pro limit, and image generation uses more of that allowance than chat. Google publishes no per-day image count for any plan, so an "N Pro images per day" figure is not a Google number. Surface-by-surface limits are in Nano Banana Pro Limits.
Google's AI Pro membership help supports monthly and annual billing for personal accounts. Google lists annual billing for AI Pro at $199.99 a year in the US, less than twelve monthly payments of $19.99 ($239.88). The checkout still owns the actual total, promotion, tax, currency and payment date, and prices outside the US differ by country.
Canceling a Google AI plan affects more than image access
For a direct Google One membership, you cancel in Google One settings with Cancel membership. Google says storage and benefits stay until the end of the billing cycle, and its Google One cancellation help adds that storage plan purchases are non-refundable in most countries or regions.
Check Drive, Gmail, Photos storage and every bundled benefit before canceling. If you subscribed on an iPhone or iPad, cancel in the App Store and ask Apple for any refund; if a carrier or another partner owns the charge, that seller owns cancellation and refunds. "Cancel anytime" does not mean "refund anytime."
Third-party per-call prices: LaoZhang API Pro at $0.09, 2.1 at $0.045
As of October 29, 2026, the LaoZhang API Nano Banana Pro documentation lists gemini-3-pro-image at $0.09 per call at 1K, 2K and 4K, with generation and editing at the same price and thinking included, through api.laozhang.ai. The model comes from Google's official channel, and you need only a LaoZhang API key, not a Google account or a Google billing project. Its Nano Banana 2.1 page lists gemini-nano-banana-2.1 at $0.045 per call from 1K to 4K. Google Search grounding adds $0.014 per search on both. These are LaoZhang API prices, not changes to Google's rates, and the full list is on LaoZhang API models and pricing.
Disclosure: laozhang.ai is our own service. Its prices are platform pricing, not an independent comparison.
Nano Banana Pro 4K price per image: $0.09 against $0.12–$0.30
At 4K, the flat call price sits below both of Google's documented rates and below two other per-call providers. Provider prices below come from the Kie.ai Nano Banana Pro page and the fal.ai Nano Banana Pro model page as of October 29, 2026.
| Nano Banana Pro at 4K | Price per image | Billing unit |
|---|---|---|
| LaoZhang API | $0.09 | Per call, same price at 1K and 2K |
| Google Gemini API, Batch | $0.12 | Per image, queued job |
| Kie.ai | about $0.12 | 24 credits per 4K image |
| Google Gemini API, Standard | $0.24 | Per image, plus input and thinking tokens |
| fal.ai | about $0.30 | $0.15 per image, 4K billed at double |
The saving against Google Standard is (0.24 − 0.09) / 0.24 = 62.5% at 4K and (0.134 − 0.09) / 0.134 = 32.8% at 1K and 2K, before Google's input and thinking tokens, which the per-call price already covers. fal.ai's $0.15 base price is also above $0.09 at 1K and 2K. For Nano Banana 2.1, $0.045 per call is below Google's Standard 2.1 price at 2K ($0.0504) and at 4K ($0.113). Exact 4K pixel sizes for each aspect ratio are in Nano Banana Pro 4K: Exact Sizes, $0.24 Cost and Watermark Rules.
LaoZhang API key limits: no per-key RPM, IPM or concurrency cap
A single LaoZhang API key has no per-key RPM (requests per minute), IPM (images per minute) or concurrency cap, and requests fail over across several official Google routes. An occasional 429 still happens when upstream capacity is short; it is not charged and can be retried after a short wait. On a Google project, by contrast, Tier 1 stops at $10 of spend per rolling 10 minutes, about 74 Standard images at 1K/2K.
On October 29, 2026, six requests were sent at the same moment from one LaoZhang API key on an ordinary customer account, through the Gemini-native generateContent endpoint: one 4K image at 16:9, one 2K image at 1:1 and four 1K images at 1:1. All six returned HTTP 200 within 70.4 seconds, with no 429 or 5xx. The 4K image took 70.3 seconds and came back at 5504×3072 pixels (an 8.2 MB JPEG), the 2K image at 2048×2048 in 42.5 seconds, and the 1K images at 1024×1024 in 23.7–27.4 seconds. The usage log showed six charges of $0.09, $0.54 in total, with the 4K image billed the same as each 1K image. A 100% crop of the 4K image shows paper-fiber texture, separate string strands and crisp label lettering. One account, one machine and six simultaneous calls are not a load test or an availability figure. Google's 4K output at 16:9 is 5504×3072, not 3840×2160.
First 4K call with gemini-3-pro-image
Create the key in the LaoZhang API console with the billing mode set to Usage first or Per-call. Nano Banana Pro has only a per-call price, so a key set to Usage-based billing can't call it. Use the Gemini-native endpoint when size matters, because it is the one that accepts imageSize (1K, 2K or 4K) and aspectRatio; the OpenAI-compatible chat completions route has no size or ratio settings.
curl --max-time 300 \
"https://api.laozhang.ai/v1beta/models/gemini-3-pro-image:generateContent" \
-H "Authorization: Bearer $LAOZHANG_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"contents": [{
"role": "user",
"parts": [{"text": "Product photo of a ceramic mug with a paper label that reads NANO BANANA PRO 4K TEST"}]
}],
"generationConfig": {
"responseModalities": ["IMAGE"],
"imageConfig": {"aspectRatio": "16:9", "imageSize": "4K"}
}
}' \
-o pro-4k.jsonThe image arrives as Base64 in an inlineData part of candidates[0].content.parts; skip any part marked thought and decode the last remaining one. A script that saves the file, plus JSON and YAML prompt formats, is in Nano Banana Pro API Guide: JSON, YAML, Image Saving, and PDFs.
The billable event differs from Google's. LaoZhang API charges every HTTP 200 response once, even when no image comes back, for example after a safety block. A non-200 response such as 400, 429 or 5xx is not charged. Check call logs before resending.
Before testing any provider, get clear answers for:
- the actual model ID and supported resolution;
- the event that creates a charge;
- empty output, safety rejection, timeout and retry treatment;
- image and log retention;
- credit expiry, refund, support and exit terms;
- an exportable call log that can be reconciled with the bill.
Stop if the model, billable event or data handling cannot be explained. Use Google directly when first-party billing or governance matters more than a lower posted call price.
For the changing consumer entitlement, read Nano Banana Pro free versus paid limits. For hands-on product access, use the Nano Banana Pro how-to guide. A free API key is a separate question covered by the API key and free-tier guide.
Which Nano Banana Pro price applies to you
- Manual use in the Gemini app: a Google AI plan, from $19.99 a month for AI Pro in the US. You get Nano Banana 2.1 or 2 by default and Pro through Redo with Pro, with no published image count. Before paying, check what the no-plan tier already gives you in Can You Use Nano Banana for Free? Google's No-Cost Options.
- Interactive API work: Standard at $0.134 per 1K/2K image and $0.24 at 4K, plus thinking and input. Reconcile the first project bill.
- Exact text in the image: Pro. For everything else, test Nano Banana 2.1 first at a quarter of the 1K price.
- Flat per-call billing: a provider such as LaoZhang API at $0.09 for Pro at 1K, 2K and 4K, tested with five outputs and its own call logs before volume.
Disclosure: laozhang.ai is our own service. Its prices are platform pricing, not an independent comparison.
When the model ID, billing unit or source page changes, drop the old number. The order that survives a price change is the same: name the invoice, name the unit, then divide the real bill by accepted outputs.
Nano Banana Pro pricing: other questions
Does Nano Banana Pro cost money?
Yes, every route that returns a Pro image is paid. As of October 29, 2026, Google's Gemini API pricing page lists "Not available" in the free-tier column for Nano Banana Pro, Nano Banana 2 and Nano Banana 2.1, and in the Gemini app Pro is a Redo with Pro option for Google AI plan subscribers. What you can get without paying is Nano Banana 2 in the Gemini app; the full picture of free access is in Is Nano Banana Pro Free? Gemini Plans, Flow Limits and the API.
How much is Nano Banana 2?
On Google's Gemini API, Nano Banana 2 (gemini-3.1-flash-image) costs $0.067 at 1K, $0.101 at 2K and $0.151 at 4K on the Standard rate as of October 29, 2026. LaoZhang API charges $0.055 per call for it at any size from 512px to 4K, generation and editing alike. Nano Banana 2 Lite and the newer 2.1 are priced side by side in Nano Banana API Pricing & Cost: 2.1, 2 Lite, Pro, No Free Tier.
Disclosure: laozhang.ai is our own service. Its prices are platform pricing, not an independent comparison.
Sources15
External pages this guide links to, in the order they appear. Last updated Nov 11, 2026.
Sources15
External pages this guide links to, in the order they appear. Last updated Nov 11, 2026.
- 1.Gemini API pricing pageai.google.dev/gemini-api/docs/pricing
- 2.Gemini API deprecations pageai.google.dev/gemini-api/docs/deprecations
- 3.Vertex AI pricing pagecloud.google.com/gemini-enterprise-agent-platform/generative-ai/pricing
- 4.Gemini API Flex inference pageai.google.dev/gemini-api/docs/flex-inference
- 5.Gemini API rate limits pageai.google.dev/gemini-api/docs/rate-limits
- 6.Gemini API billing pageai.google.dev/gemini-api/docs/billing
- 7.US Google AI plans pagegemini.google/us/subscriptions
- 8.Google AI plans pageone.google.com/about/google-ai-plans
- 9.Gemini image generation overviewgemini.google/overview/image-generation
- 10.Gemini Apps image help pagesupport.google.com/gemini/answer/14286560
- 11.Gemini Apps limits pagesupport.google.com/gemini/answer/16275805
- 12.AI Pro membership helpsupport.google.com/googleone/answer/16476811
- 13.Google One cancellation helpsupport.google.com/googleone/answer/9056360
- 14.Kie.ai Nano Banana Pro pagekie.ai/nano-banana-pro
- 15.fal.ai Nano Banana Pro model pagefal.ai/models/fal-ai/nano-banana-pro





