As of July 2, 2026, the official Nano Banana Pro API price belongs to Google's gemini-3-pro-image row: $0.134 for a 1K or 2K output and $0.24 for a 4K output in Standard mode, with no official Free Tier on that Pro image API row.
Batch/Flex is the official lower-cost lane for workloads that can wait, at $0.067 for 1K/2K and $0.12 for 4K. If the job is price-sensitive before it is Pro-sensitive, compare Nano Banana 2 and Nano Banana 2 Lite as separate official model rows; if the route is a provider or gateway, treat the price as that provider's contract, not Google's official API price.
| Route / mode | Current price signal checked on July 2, 2026 | Free Tier status | Owner | First use |
|---|---|---|---|---|
| Nano Banana Pro Standard | $0.134 for 1K/2K; $0.24 for 4K | Not available | Google official API, gemini-3-pro-image | Interactive Pro-quality budgeting |
| Nano Banana Pro Batch/Flex | $0.067 for 1K/2K; $0.12 for 4K | Not available | Google official API, gemini-3-pro-image | Bulk or delay-tolerant jobs |
| Nano Banana 2 | Lower official rows by size | Not available | Google official API, gemini-3.1-flash-image | Cheaper official alternative before Pro is required |
| Nano Banana 2 Lite | $0.0336 Standard 1K; $0.0168 Batch 1K | Not available | Google official API, gemini-3.1-flash-lite-image | Lowest official Nano Banana-family lane when quality fits |
| Provider, app, subscription, or gateway route | Separate terms | Separate terms | That route owner | Validate model, billing, limits, refunds, and logs before relying on it |
Use the Google API rows for developer budgeting first. Use provider prices, app quotas, subscriptions, or credits only after the route owner is clear enough to explain who bills the call, who owns limits, and what happens to failed or retried generations.
Quick Price Ladder for the Official API
The official cost-per-image answer starts with the Google AI for Developers pricing page, because that page owns the Gemini API price rows. In that contract, Nano Banana Pro is the market name readers attach to gemini-3-pro-image; the row is an output price for that model route, not a universal Nano Banana price across every app, subscription, wrapper, or provider dashboard.
The Pro ladder has two practical rules. First, Standard mode groups 1K and 2K output together at $0.134, then prices 4K output separately at $0.24. Second, Batch/Flex lowers the same Pro output lane to $0.067 for 1K or 2K and $0.12 for 4K. A developer doing asynchronous or delay-tolerant work should not budget from the Standard row just because Standard is the easiest row to read.
| Official Pro API lane | Model route | 1K output | 2K output | 4K output | Best use |
|---|---|---|---|---|---|
| Standard | gemini-3-pro-image | $0.134 | $0.134 | $0.24 | Interactive generation, direct API tests, latency-sensitive calls |
| Batch/Flex | gemini-3-pro-image | $0.067 | $0.067 | $0.12 | Bulk jobs that can accept batch or flexible execution |
The image generation docs add the request-shape boundary: Gemini 3 image generation defaults to 1K, and 2K or 4K must be requested with supported output-size controls such as 2K and 4K. The pricing page answers the budget question; the image docs answer whether the request actually asks for the output size you plan to pay for.
What the Official Price Includes and Excludes
The visible price row is an output baseline. It is useful because it gives a clean number for a successful output on one official API lane, but it is not the same as every invoice, every retry path, or every third-party billing rule.
Image input is separate. Google currently lists the Gemini 3 Pro Image input image cost as 560 tokens, equivalent to $0.0011 per image input. If every generation uses one reference image, that adds about $0.11 for 100 image inputs or $1.10 for 1,000 image inputs before counting output. A workflow that sends multiple reference images per call changes that input component again.
Grounding and extra API behavior also need their own checks. Google Search grounding is not the same cost shape as a plain image output call, and project limits or rate behavior are account and model specific. Edits, retries, failed attempts, queueing, and accepted-output rules can change again when the route is a provider. The official output row tells you what a successful Pro image output costs on Google's API lane; it does not prove how a provider handles failed generations, credits, refunds, or support.
For a clean internal estimate, split the budget into four lines:
| Budget line | Who owns it | How to treat it |
|---|---|---|
| Pro output price | Google official API pricing | Use the 1K/2K/4K Standard or Batch/Flex row with the checked date. |
| Image input | Google official API pricing | Add $0.0011 per input image when the workflow sends references. |
| Grounding and special API behavior | Google official API pricing and docs | Recheck before using grounded requests or special features. |
| Subscription, app, provider, or gateway terms | That surface or provider | Do not fold into official API cost unless the contract explicitly says it is API billing. |
Free Tier Boundary: Key, App, Subscription, or Provider Credit
The Free Tier answer is where most Nano Banana Pro API pricing pages become unsafe. An API key can be created without proving that the Pro image API row is free. A Gemini app quota can let a person manually test images without becoming a Developer API billing entitlement. A subscription can control consumer access without defining API output cost. Provider credits can lower purchase friction without changing Google's official price row.
Use this split before trusting any "free Nano Banana Pro API" claim:
| Surface | What it can mean | What it does not prove |
|---|---|---|
| API key | A credential associated with a project, billing setup, quota, and authorization rules | A free Pro image output entitlement |
| Developer API pricing row | Official model, output price, Free Tier status, and billing owner | Consumer app quotas or provider credits |
| Gemini app or consumer plan | Manual app access, app-surface limits, export behavior, plan eligibility | Official API cost per image |
| Subscription or partner surface | That surface's monthly plan, credit pool, reset cadence, export rules | Google-wide API usage or universal Pro entitlement |
| Provider or gateway credit | That provider's dashboard balance, model route, logs, support, and refund terms | Google's official price row or another provider's contract |
For official API budgeting, the Pro image API row says Free Tier is not available. If a page says Nano Banana Pro is free, ask which owner it means before using the number: app quota, subscription benefit, trial credit, provider balance, or a real official API pricing row. Only the last one can answer official Developer API cost per image.
Official Cheaper Alternatives: Batch/Flex, Nano Banana 2, and Lite
Cheaper alternatives exist, but they are not all cheaper Pro prices. Batch/Flex is still Pro, still official, and still gemini-3-pro-image; it is cheaper because the execution mode is different. Nano Banana 2 and Nano Banana 2 Lite are different official model lanes, so they belong in an alternatives decision, not in the Pro row.
As checked on July 2, 2026, the official lower-cost rows look like this:
| Alternative lane | Official model ID | Standard output price | Batch/Flex or Batch output price | How to use the number |
|---|---|---|---|---|
| Nano Banana Pro Batch/Flex | gemini-3-pro-image | Use Standard only if needed | $0.067 for 1K/2K; $0.12 for 4K | Best official Pro discount when latency is flexible |
| Nano Banana 2 | gemini-3.1-flash-image | $0.045 for 0.5K; $0.067 for 1K; $0.101 for 2K; $0.151 for 4K | $0.022, $0.034, $0.050, $0.076 by the same sizes | Lower-cost official image lane, not a Pro price |
| Nano Banana 2 Lite | gemini-3.1-flash-lite-image | $0.0336 for 1K | $0.0168 for 1K | Lowest official Nano Banana-family lane when quality and features fit |
That model split is the article's main cost decision. Use Pro when the job depends on Pro capability, higher-fidelity review, or the specific Pro model contract. Use Nano Banana 2 when the job can start with a lower official row and escalate only if output quality or reference fidelity fails. Use Lite only when a basic 1K route is acceptable and the lower price matters more than Pro-level capability. None of those choices remove the need to label owner, mode, resolution, and checked date beside every price.
Cost Calculator for 10, 100, and 1,000 Outputs
Once the route is clear, the output math is simple: multiply the official output price by accepted outputs, then add input, grounding, retry, and route-specific costs separately.
| Pro API mode | Output size | 10 outputs | 100 outputs | 1,000 outputs | What is excluded |
|---|---|---|---|---|---|
| Standard | 1K | $1.34 | $13.40 | $134.00 | Input images, grounding, retries, app/provider terms |
| Standard | 2K | $1.34 | $13.40 | $134.00 | Input images, grounding, retries, app/provider terms |
| Standard | 4K | $2.40 | $24.00 | $240.00 | Input images, grounding, retries, app/provider terms |
| Batch/Flex | 1K | $0.67 | $6.70 | $67.00 | Input images, grounding, retries, app/provider terms |
| Batch/Flex | 2K | $0.67 | $6.70 | $67.00 | Input images, grounding, retries, app/provider terms |
| Batch/Flex | 4K | $1.20 | $12.00 | $120.00 | Input images, grounding, retries, app/provider terms |
For example, 1,000 Standard 4K Pro outputs cost $240.00 in output fees before any reference-image input or special route behavior. If those 1,000 requests each use one image input, add about $1.10 for that input component. If the workflow must run interactively and cannot use Batch/Flex, do not apply the cheaper Batch/Flex row just because it is visible.
For exact implementation details such as output-size request shape and 4K verification, use the Gemini 4K image generation API guide. Keep implementation testing separate from price math so a misconfigured request does not look like a pricing discrepancy.
When a Provider Price Belongs in a Separate Decision
Provider pages often answer a different job: local payment, OpenAI-compatible endpoints, unified logs, support, lower visible credit cost, or an easier dashboard. Those can be legitimate reasons to test a route, but they are not official Google API pricing.
Use the official cost ladder first, then move provider evaluation into a proof checklist:
| Provider proof item | Evidence to collect before relying on it |
|---|---|
| Model route | Does the dashboard, request log, or accepted output show the actual model route? |
| Accepted-output cost | Are failed generations billed, refunded, retried, or hidden inside credits? |
| Throughput and limits | What rate behavior appears in a real workload, not just a price card? |
| Support and billing | Are invoices, orders, logs, and support channels usable for production review? |
| Data and rights | What happens to uploaded references, stored outputs, commercial use, and deletion? |
For that provider-specific job, use the Nano Banana Pro API route validation guide after the official baseline is clear. The provider question is "is this route worth testing for my workload?", not "what is Google's official Pro cost per image?"
Practical Budget Scenarios
If you are running a small editorial or design test, start with the Standard output row and count accepted outputs. Ten 2K Pro outputs cost about $1.34 in output fees before inputs or extra features. That is usually enough to learn whether Pro is worth the higher row before building automation.
If you are preparing a bulk asset refresh, compare Standard with Batch/Flex early. A 1,000-image 2K job is $134.00 in Standard output fees or $67.00 in Batch/Flex output fees, before exclusions. That difference is large enough that scheduling tolerance becomes a pricing decision.
If the workload is high-resolution production, treat 4K as its own budget lane. A 1,000-image 4K job is $240.00 Standard or $120.00 Batch/Flex in output fees. Add input images, revision passes, and review rejects separately. A 4K workflow with repeated reference uploads can be more expensive than the output row suggests, even when the output price is correct.
If the real question is model choice, compare Pro against Nano Banana 2, Lite, and other image routes before buying volume. The GPT Image 2 vs Nano Banana Pro route comparison is a better place for cross-model tradeoffs once the official Pro price is no longer the only decision.
Checklist Before You Use a Number
Before copying any Nano Banana price into a budget, verify five fields:
- Model:
gemini-3-pro-imagefor Nano Banana Pro,gemini-3.1-flash-imagefor Nano Banana 2, andgemini-3.1-flash-lite-imagefor Nano Banana 2 Lite. - Mode: Standard, Batch/Flex, or Batch.
- Resolution: 0.5K, 1K, 2K, or 4K where that model supports it.
- Price scope: output-only, plus any image input, grounding, retry, or feature cost.
- Route owner: Google official API, Gemini app, partner surface, subscription, provider, or gateway.
If any field is missing, the number may still be a useful clue, but it is not safe enough for production budgeting. The most common mistake is treating the cheapest visible Nano Banana number as the official Pro price. The safer habit is to name the model, owner, mode, resolution, and checked date beside every price.
FAQ
How much does Nano Banana Pro cost per image in 2026?
As checked on July 2, 2026, official Nano Banana Pro API output pricing is $0.134 for a 1K or 2K image and $0.24 for a 4K image in Standard mode. Batch/Flex lowers those output prices to $0.067 for 1K or 2K and $0.12 for 4K. Those are output prices for gemini-3-pro-image, not all-in subscription, app, or provider costs.
Is there a Nano Banana Pro API Free Tier?
For the official Pro image API row checked on July 2, 2026, Free Tier is not available. Do not treat API key creation, Gemini app quota, subscription access, trial credits, or provider credits as proof of a free official Pro image API entitlement.
Why do 1K and 2K have the same Pro price?
Google's current Pro pricing row groups 1K and 2K together for gemini-3-pro-image output. Do not infer that every model behaves that way: Nano Banana 2 has its own price ladder, Lite has a 1K lane, and 4K Pro output has a higher row.
Is 4K Nano Banana Pro more expensive?
Yes for the official Pro API output row. Standard 4K output is $0.24 per image, while Standard 1K or 2K output is $0.134. Batch/Flex 4K output is $0.12, compared with $0.067 for 1K or 2K.
Does Batch/Flex change the official price?
Yes. Batch/Flex is an official Google API mode, and the checked Pro output rows are lower than Standard: $0.067 for 1K or 2K and $0.12 for 4K. Use it only when the workload can accept the execution model; do not apply the cheaper row to interactive calls that require Standard behavior.
Is Nano Banana 2 the same pricing as Nano Banana Pro?
No. Nano Banana 2 maps to gemini-3.1-flash-image; Nano Banana Pro maps to gemini-3-pro-image. Nano Banana 2 has lower official rows in the current pricing page, but those rows should not be used as Pro prices.
What is the cheapest official Nano Banana-family API lane?
As checked on July 2, 2026, Nano Banana 2 Lite has the lowest official row in the Nano Banana-family lanes above: $0.0336 for a Standard 1K output and $0.0168 for Batch. That is not a Pro price, and it should only be used when the Lite model's quality and feature boundary fits the job.
Does a Gemini or Google AI subscription include API cost per image?
Do not assume that. Consumer plans and Gemini app access are app-surface contracts. Official API cost per image is governed by Google AI for Developers pricing, billing, model availability, and project limits.
Can a provider offer a cheaper Nano Banana Pro price?
A provider can offer its own route, credits, dashboard price, support, and billing terms. That does not change Google's official API price. Treat provider prices as provider contracts and verify model route, accepted-output billing, failed-job handling, logs, support, data terms, and rights before relying on them.
Which model ID should developers use for Nano Banana Pro?
Use gemini-3-pro-image for the current official Pro API lane. Use gemini-3.1-flash-image for Nano Banana 2 and gemini-3.1-flash-lite-image for Nano Banana 2 Lite. Avoid treating stale preview-era IDs as current unless you are explicitly reviewing historical behavior.
