Skip to main content
AI image generator and editor
Google
Nano Banana

Nano Banana 2

Kyeo AI offers generation and reference-guided editing through `nano-banana-2`, with optional `google_search`, up to 14 `image_input` files, 15 aspect-ratio choices, and 1K, 2K, or 4K output. Those tiers currently cost 5, 8, and 12 credits, and prompts allow 20,000 characters.

Capabilities
Text-to-Image and Image-to-Image
Credit cost
5–12 credits per request
Prompt limit
20,000 characters
Upload limit
Up to 14 images
Text-only requests generate a new image; uploaded references stay on `nano-banana-2` and are sent through `image_input`.
Prompts allow 20,000 characters, and each request accepts up to 14 JPG, PNG, or WebP images.
`google_search` can use Google Web Search information, but the resulting image is not a substitute for fact-checking.
Kyeo currently exposes only 1K, 2K, and 4K, priced at 5, 8, and 12 credits.
30-second overview
Nano Banana
5–12 credits per request
What it does best
Among Kyeo's current Nano Banana options, this page is distinguished by a Google Web Search toggle, up to 14 images, a 20,000-character prompt, 15 aspect-ratio choices, and 1K, 2K, or 4K output tiers.
Best for
It is a practical candidate for visuals that need recent public information as context, edits built from several clearly assigned references, or infographics and covers with detailed text and layout constraints. Search results, in-image text, and subject details still need review.
Popular searches
How do I use Nano Banana 2 on Kyeo AIHow many credits do Nano Banana 2's 1K, 2K, and 4K tiers costWhen should I enable Google Search in Nano Banana 2

At a glance

What is this model like?

Google calls Nano Banana 2 Gemini 3.1 Flash Image and positions it as an efficient all-purpose image model. Upstream interface uses its own `nano-banana-2` identifier without publishing the upstream Gemini version on the endpoint page. This page therefore uses Google material for model background and upstream interface plus Kyeo configuration for the controls users can actually select.

What it does best
Among Kyeo's current Nano Banana options, this page is distinguished by a Google Web Search toggle, up to 14 images, a 20,000-character prompt, 15 aspect-ratio choices, and 1K, 2K, or 4K output tiers.
Best for
It is a practical candidate for visuals that need recent public information as context, edits built from several clearly assigned references, or infographics and covers with detailed text and layout constraints. Search results, in-image text, and subject details still need review.
Why use it on Kyeo AI
Search, aspect ratio, resolution, and output format are visible in one form, and Kyeo calculates credits from the chosen resolution. You can validate composition and instruction following at 1K before paying for 2K or 4K.

Key facts

Quickly assess whether this model matches your workflow.

Category
AI image generator and editor
Developer
Google
Model family
Nano Banana
Integration
Current site integration
Platform model ID
nano-banana-2
Runtime
Asynchronous task
Prompt limit
20,000 characters
Upload limit
Up to 14 images

Also known as

The same model may appear under different names across documentation and community discussions; this list keeps them easy to search and compare.

nano-banana-2
Gemini 3.1 Flash Image
gemini-3.1-flash-image
Google Nano Banana 2
Nano Banana 2 API

Common search intents

These questions and keywords combine public references with common creative use cases to help you choose faster.

How do I use Nano Banana 2 on Kyeo AI
How many credits do Nano Banana 2's 1K, 2K, and 4K tiers cost
When should I enable Google Search in Nano Banana 2
Which image formats and how many files does Nano Banana 2 accept
Is 20,000 characters a recommended Nano Banana 2 prompt length
How should I choose a Nano Banana 2 resolution tier
How does 3-credit Nano Banana compare with Nano Banana 2
Which controls differ between Nano Banana 2 and Pro
How do I edit with reference images in Nano Banana 2
How should I fact-check an infographic made with Nano Banana 2

Selection guide

Use these decision points when choosing a model.

1
Choose it when you need search, more references, or resolution tiers

Nano Banana 2 is worth testing when the brief depends on recent public information, more references than the base route's recommended working set, extreme 4:1, 1:4, 8:1, or 1:8 ratios, or explicit 2K or 4K output.

2
Compare other models by controls and test output

For basic generation or editing at a fixed 3 credits, compare standard Nano Banana. When search is unnecessary and 8 images are enough, run the same prompt and material through Pro.

3
Build a 1K baseline with a small reference set

Begin with `google_search` off, 1 to 3 essential images, and a 1K request. Enable search for a second run only when the brief needs recent information. Raise resolution or add inputs after the source roles and composition are working.

Popular comparisons

Common comparison questions help you find the practical differences quickly.

What are the tradeoffs between fixed-price Nano Banana and tiered Nano Banana 2
When are standard Nano Banana's current controls already enough
Do both Nano Banana 2 and Pro expose a Google Search control
What do 14-image and 8-image reference limits mean in practice
How do Nano Banana 2 and GPT Image 1.5 compare on cost and input limits
How should I test different image models with the same brief and source material

Model comparison

Compare the current model with alternatives at a glance.

Dimension
Nano Banana 2
Nano Banana
Nano Banana Pro
Current Kyeo controls
Search toggle, 14 images, 15 aspect-ratio choices, and three output tiers
Fixed at 3 credits, with no resolution tier or search control
Three resolution tiers, up to 8 images, and no search control
Request structure
The same `nano-banana-2` ID accepts text and optional `image_input`
Kyeo switches model IDs based on whether images are attached
The same `nano-banana-pro` ID accepts text and images
Prompt and image limits
20,000 characters and up to 14 input images
5,000 characters; the editing route accepts up to 10 images
10,000 characters and up to 8 images
Current Kyeo price
5 credits at 1K; 8 at 2K; 12 at 4K
3 credits for either generation or editing
8 credits at 1K or 2K; 14 at 4K
Verifiable distinction
The only one of these Kyeo options with a Google Search toggle, plus up to 14 images
Fixed-price generation and editing with fewer controls
OpenAI image generation and editing, currently with up to 16 references

Practical usage insights

Practical observations based on real creative workflows.

A shared product name does not disclose the upstream service version

Google maps Nano Banana 2 to stable `gemini-3.1-flash-image`, while the upstream interface publishes `nano-banana-2`.

Direct Gemini features and Kyeo controls are different scopes

Google's model also supports 0.5K, image-search grounding, PDF or video context, and Thinking.

Credit differences are not a quality ranking

Nano Banana 2 currently costs 12 credits at 4K and Pro costs 14 at 4K. That is a Kyeo pricing difference, not evidence about image quality, text accuracy, reference adherence, or editing stability; compare those with the same source material.

Capabilities

Text generation and reference-guided editing

Kyeo submits both text-only and image-guided requests to `nano-banana-2`. Uploaded references are passed through `image_input`, with a maximum of 14; that limit is not a recommendation to fill every slot.

Optional Google Web Search

`google_search` lets generation use Google Web Search information for time-sensitive topics such as weather, sports, or recent events. It does not attach verifiable citations to the image or guarantee that every word, number, or graphic is correct.

Aspect ratio, resolution, and output format

The form has 15 aspect-ratio choices, including auto, 4:1, 1:4, 8:1, and 1:8. Output options are 1K, 2K, or 4K and JPG or PNG; Google's 0.5K tier is not exposed on Kyeo.

Prompt and credit boundaries

Prompts allow 20,000 characters, but that is a ceiling rather than a target. A request currently costs 5 credits at 1K, 8 at 2K, or 12 at 4K, so it is sensible to validate instructions at the lower tier first.

Use cases

Drafts informed by recent public information

Enable `google_search` for weather, sports, or recent-event context. Check every date, number, map, chart, and piece of text against a reliable source before using the output.

Product or brand visuals with several references

Assign separate images to the subject, material, setting, and layout. When identity, product accuracy, or brand rules matter, inspect details that were not meant to change.

Infographics or covers with text and layout constraints

Break out the title, hierarchy, composition, palette, and protected elements. Google reports better multilingual text rendering for this model, but Kyeo output still requires line-by-line proofreading.

Prompt tips

Decide whether search is actually necessary

Keep `google_search` off for historical, fictional, or style-only work. Turn it on when the image needs recent public information, then verify against the original sources instead of treating the generated image as evidence.

Start with a small, clearly assigned image set

Begin with 1 to 3 images and identify which one supplies the subject, material, pose, setting, or layout.

Turn a long prompt into testable sections

The 20,000-character allowance is a limit, not a goal. Separate subject, composition, copy, reference roles, protected details, and exclusions, then test at 1K before choosing a higher output tier.

Strengths

Supports both text-to-image generation and reference-image editing on one page.
Kyeo exposes a Google Web Search toggle for time-sensitive subjects.
The form offers 15 aspect-ratio choices and 1K, 2K, or 4K output.
A 20,000-character prompt and up to 14 images leave room for complex constraints.

Limitations

14 images is the upstream interface maximum, not a recommended input count or a consistency guarantee.
`google_search` can introduce irrelevant or incorrect material, and a generated image does not include fact-checkable citations.
The 20,000-character allowance is only a length limit. Longer instructions can still contradict one another. Technical identifiers: `google_search`.
Kyeo does not expose the direct API's 0.5K tier, image-search grounding, PDF or video input, or Thinking controls.

Adjustable parameters

Quickly assess whether this model matches your workflow.

Aspect ratio
aspect_ratio
Optional
Parameter type: Select
Default: auto
Auto
1:1
16:9
9:16
3:2
2:3
15 choices
Resolution
resolution
Optional
Parameter type: Select
Default: 1K
1K
2K
4K
Output format
output_format
Optional
Parameter type: Select
Default: jpg
JPG
PNG
Google Search grounding
google_search
Optional
Parameter type: Select
Default: Off
Off
On

Credit usage

5–12 credits per request

Current Kyeo AI pricing is 5 credits at 1K, 8 credits at 2K, and 12 credits at 4K.

Budget tip

Before batch generation, run an A/B test with the same assets across the current model and alternatives to validate quality and avoid wasting credits.

FAQ

Related models

Compare these similar candidates before deciding.

Nano Banana

Nano Banana, also known as Gemini 2.5 Flash Image, provides text-to-image generation and natural-language editing on Kyeo AI. Requests without images use `google/nano-banana`; requests with images use `google/nano-banana-edit`. Both paths cost 3 credits and accept prompts up to 5,000 characters.

AI image generator and editor
3 credits per request

Nano Banana Pro

Nano Banana Pro supports text-to-image generation and reference-image editing on Kyeo AI, with 1K, 2K, or 4K output. Each 1K or 2K request costs 8 credits, while 4K costs 14. Resolution is an output-size tier, not a guarantee of quality, accuracy, or detail.

AI image generation and editing model
8–14 credits per request

GPT Image 1.5

GPT Image 1.5 supports generation and editing on Kyeo AI. Text-only requests use `gpt-image/1.5-text-to-image`; attaching references switches to `gpt-image/1.5-image-to-image`. Kyeo exposes 3 aspect ratios and `medium / high` tiers at 4 / 22 credits. OpenAI now marks GPT Image 1.5 as a deprecated previous model, so new work should also benchmark a current model such as GPT Image 2.

AI image model
4-22 credits / request

Sources

Content is based on model documentation, feature references, and the settings available on Kyeo AI.

Source notes

This page uses Kyeo's current configuration and generation route, the upstream interface's Nano Banana 2 schema, and Google's official Gemini 3.1 Flash Image material. Google maps Nano Banana 2 to `gemini-3.1-flash-image`; upstream interface publishes its own `nano-banana-2` identifier without naming the upstream Gemini version. Kyeo exposes a 20,000-character prompt, up to 14 images, Google Web Search, 15 aspect-ratio choices, 1K, 2K, or 4K output, and JPG or PNG. Other direct Gemini API capabilities, including 0.5K, image-search grounding, PDF or video input, and Thinking controls, are not treated as available on Kyeo.

Last reviewed: 2026-07-20
Google Blog: Nano Banana 2 launch
Official
View source
Google Blog: Build with Nano Banana 2
Official
View source
Google AI for Developers: Gemini 3.1 Flash Image
Official
View source
Google AI for Developers: Image generation with Gemini
Official
View source
Upstream interface documentation: Nano Banana 2
Platform

The upstream interface supports up to 14 JPG, PNG, or WebP `image_input` files at 30 MB each, `google_search`, 15 `aspect_ratio` choices, 1K, 2K, or 4K `resolution`, `output_format`, and prompts up to 20,000 characters.

Current Kyeo credit configuration
Platform

Kyeo currently charges 5, 8, or 12 credits for 1K, 2K, or 4K respectively.

Kyeo AI model configuration
Platform

Kyeo uses the upstream `nano-banana-2` identifier for both text and image requests, accepts images through optional `image_input`, and shows only the controls supported by the current form.