Nano Banana Pro API
Nano Banana Pro is Google's officially named Gemini 3 Pro Image model for image generation and editing, built on Gemini 3 Pro reasoning. It targets complex infographics, multilingual in-image text, and high-resolution output. Roko exposes it through two async task routes billed per image and resolution tier.
- Public model ID
nano-banana-pro- Output tiers
1K / 2K / 4K- Unit price
from $0.018/image- Generation endpoint
POST /v1/images/tasks- Catalog snapshot
What is Nano Banana Pro?
Nano Banana Pro is the image generation and editing model Google DeepMind released on November 20, 2025, officially named Gemini 3 Pro Image and built on Gemini 3 Pro. Official materials emphasize three points: reasoning handles complex multi-turn generation and editing; Google Search grounding can produce fact-based infographics; and clear multilingual text renders directly inside images. The official Vertex spec also lists up to 14 reference images, person consistency for up to 5 people, and 1K/2K/4K resolutions.
Roko exposes it as nano-banana-pro with two routes: text-to-image and image-to-image (image_urls accepts URLs or base64). There are 11 aspect ratios (auto default) and three resolution tiers at $0.018, $0.024, and $0.036 per image, with batch n scaling linearly. Reference-image limits follow Roko request validation; the official 14-image spec is upstream capability that this site has not verified, so it is not promised on the page.
Capability and quality claims all cite Google official materials with sources; callable parameters and prices follow the current Roko public catalog. Google's own API pricing differs and is neither quoted nor converted here. All quality demonstrations for this model are treated as not-Roko-measured; see the case-baseline note below.
Key capabilities
Reasoning-based complex generation
Google positions Gemini 3 Pro reasoning for image tasks: educational infographics, step-by-step diagrams, and fact-based charts rather than single-shot output. Quality claims follow official materials and must be validated on your own prompts.
Multilingual in-image text
Official materials highlight clear, legible multilingual text rendering — taglines, long paragraphs, calligraphy, and font textures — for posters and localized assets.
1K/2K/4K resolutions
Roko's catalog exposes three tiers at $0.018, $0.024, and $0.036 per image. 4K suits print and large-format; prototype at 1K first.
Image-to-image with references
image_urls accepts URLs or base64 references for product composites, style transfer, and person consistency; limits follow request validation.
Nano Banana Pro API pricing
Prices come from the current Roko model catalog, billed per output image and resolution tier. Confirm the console estimate before submitting tasks.
| Resolution | Price | Billing unit |
|---|---|---|
| 1k | $0.018 | per output |
| 2k | $0.024 | per output |
| 4k | $0.036 | per output |
How costs add up
The standard cost for prototypes and everyday artwork.
A typical batch for picking the best take from one task.
The final-output tier for print or large-format display.
Parameters and supported range
Fields come from the current Roko public generation_config: 11 aspect ratios (auto default), 1K/2K/4K resolutions, and both generation routes; anything else follows request validation.
| Parameter | Values | Notes |
|---|---|---|
prompt | textarea | Describe the image to create or the changes to apply to references. · Required |
image_urls | url / base64 | Reference images may use URL or Base64. State what must stay and what may change. · Optional |
aspect_ratio | auto / 1:1 / 16:9 / 9:16 / 4:3 / 3:4 / 3:2 / 2:3 / 5:4 / 4:5 / 21:9 | Selects the output aspect ratio; auto lets the model decide from the content. · Default: auto · Optional |
resolution | 1K / 2K / 4K | Selects an output resolution tier configured for this model; the value may directly affect price. · Default: 1K · Optional |
Call Nano Banana Pro with the Roko API
Use the async task flow: submit for a task_id, then poll for the result. Never expose API keys in frontend code.
Start an image generation
curl https://api.rokoapi.com/v1/images/tasks \
-H "Authorization: Bearer $API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: YOUR_UNIQUE_REQUEST_ID" \
-d '{
"model": "nano-banana-pro",
"prompt": "A clean infographic explaining how espresso is made, with labeled steps, icons and short English captions, flat vector style",
"aspect_ratio": "16:9",
"resolution": "2K"
}'Poll the task
curl https://api.rokoapi.com/v1/tasks/$TASK_ID \
-H "Authorization: Bearer $API_KEY"POST /v1/images/tasks returns 202 with a task_id; query GET /v1/tasks/{task_id} for results or failure reasons. Idempotency-Key must be unique; replays with the same key return the original task without double billing.
Where it fits
Infographics and step diagrams
The officially promoted reasoning-based composition turns knowledge into labeled visuals.
Create a step-by-step infographic for brewing pour-over coffee, with labeled stages
Multilingual posters
In-image multilingual text suits localized marketing assets.
A movie-style poster with the Japanese title text rendered clearly at the bottom
Product and brand composites
Image-to-image fuses logos, product shots, and scene references.
Place this product on a marble counter with soft morning light, keep the logo unchanged
4K masters
The 4K tier is the final-output choice for print and large-format display.
A highly detailed botanical illustration suitable for large-format printing
Limits to know before you start
- Reference-image limits follow Roko request validation; the official Vertex spec of 14 images is upstream capability, unverified on this site and not promised here.
- Search grounding (real-time facts) is an upstream capability; Roko's catalog exposes no separate toggle, so whether outputs use live information depends on the task result.
- 4K costs twice the 1K rate; batch tasks scale linearly per image — validate prompts at 1K first.
- Official outputs embed SynthID watermarks and C2PA content credentials; clear copyright, likeness, and compliance reviews before publishing.
- A successful async submit does not mean a successful generation; the task query result governs.
Nano Banana 2, 2.1, or Pro?
Pro targets reasoning-based complex generation with full 1K/2K/4K tiers (from $0.018); Lite is the low-price 1K batch tier; for efficient everyday generation and editing across 1K–4K, compare 2 and 2.1. Upstream identity and lifecycle follow the live catalog.
All models →Nano Banana Pro FAQ
What is the Nano Banana Pro API model ID?
Roko exposes nano-banana-pro. Google's official name is Gemini 3 Pro Image (model ID gemini-3-pro-image), which differs from the Roko ID — pass nano-banana-pro exactly as shown.
How much does the Nano Banana Pro API cost?
Current catalog rates: $0.018 per 1K image, $0.024 per 2K, $0.036 per 4K, with batches scaling per image. Google's official API pricing differs and is not converted here; rates may change — defer to the pre-submit estimate and settlement records.
How does it differ from Nano Banana 2 and 2.1?
Pro maps to Gemini 3 Pro Image with reasoning-heavy generation, multilingual text rendering, and full 1K/2K/4K tiers; 2 and 2.1 trace to Flash-line upstreams aimed at efficient everyday generation and editing. Choose by resolution tiers, unit price, and task type; quality differences require your own tests.
How many reference images are allowed?
The Roko catalog marks references as optional image_urls (URL or base64) with limits enforced by request validation. The official Vertex spec allows up to 14; unverified on this site and not promised here.
Does it render text accurately?
Official materials list clear multilingual text rendering as a core capability, from taglines to long paragraphs. Real accuracy varies by language, font, and layout — test in your target language before production use.
Which aspect ratios are supported?
Eleven: auto (default), 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 5:4, 4:5, and 21:9. Subject to the current Roko catalog.
Which endpoints are available?
Both the standard async image task (POST /v1/images/tasks) and the OpenAI-compatible endpoint work; this page follows the async task flow.
Can the outputs be used commercially?
Commercial suitability depends on input rights, generated content, Google's and Roko's terms, and applicable law. Official outputs embed SynthID watermarks and C2PA content credentials; clear copyright, likeness, and compliance reviews before publishing.