Annual Plan SavingsSave 50% with an annual plan
View plans
Nano Banana Pro
HomeHome
Nano Banana Series
Nano Banana 2.1
RECOMMENDED

Sharper visuals · Accurate text · Consistent subjects

Nano Banana Pro

Generate high-quality images from text descriptions

Nano Banana 2

Next-gen · Google Search · 14 Input Images

Nano Banana 2 Lite

Lightweight and fast version of Nano Banana 2

GPT Image Series
GPT Image 2.6
RECOMMENDED

Generate images, edit with references, or create GIFs with GPT Image 2.6.

In use
GPT Image 2.5

Natural lighting · Rich textures · Fine detail

GPT Image 2

Photoreal quality, precise text, pixel-level editing

GPT Image 1.5

Latest GPT 5.2 powered image generation

Seedream SeriesRELAXED
Seedream 5 Pro
NEW

Rivals Nano Banana 2, precise editing with layer control

Seedream 5 Lite

Latest Seedream model, 4K quality, 8 aspect ratios

Seedream 4.0

4K support, relaxed content policy, batch generation

Qwen Image
Qwen Image 3.0

Complex layouts · Clear text rendering · Fine realism

Creative Tools
Image Enhancement

Enhance image resolution, refine details

E-commerce Image Suite

Upload a product image and generate a complete set for major marketplaces

Nano Banana Pro Combo

Generate multi-card grid sheets in one call

GPT Image 2 Combo

Multi-card grid sheets powered by GPT Image 2

Z-Image
Z-Image

Fast photorealistic generation with relaxed content policy

Seedance Series
Seedance 2.5
NEW

30-second 4K video generation with native audio and advanced controls

Seedance 2
HOT

Cinema-quality video with native audio, up to 15s, text & image to video

Seedance 2 Mini
HOT

High-value Seedance 2 with native audio, 4-15s, 720p/1080p

Seedance 1.5 Pro

Supports first/last frame, text-to-video/image-to-video modes, cost-effective

Kling Series
Kling 4.0
NEW

Keyframes and omni reference, up to 30 seconds with native audio

Kling 3
HOT

Cinematic 4K generation with native audio

Kling o3

Omni model with first/last-frame control

Kling 2.6

Cinematic quality with native audio generation

Wan Series
Wan 3.0
NEW

Standard and high-speed Prime in one workspace

Wan 2.7
HOT

Text, image, reference, and video edit modes

Wan 2.6

Relaxed content policy

Wan 2.5

Relaxed content policy with high creative freedom

Google Series
Gemini Omni

Multimodal input × native audio, cinematic-level controllable video

Veo 3.1

Supports first/last frame video, multi-image reference video generation

Grok Series
Grok Imagine 1.5

Image-to-video by xAI Grok, flexible aspect ratios, up to 15s

Grok

Faster generation, weaker prompt adherence, supports spicy mode

Hailuo AI Series
H3 Max / Turbo

A cost-efficient MiniMax H3 Max studio for text-to-video, image-to-video and reference-led creation with native audio.

MiniMax H3 (Hailuo 03)

Hailuo 03 with text, first/last frame, and multimodal reference video modes

Hailuo AI

MiniMax Hailuo video: text/image to video, excellent first/last frame effects

OmniHuman 1.5
OmniHuman 1.5

Single image and audio to digital human video

V2V Lip Sync

Sync an existing video mouth motion to target audio

AI Chat
AI Audio
NEW

Generate dialogue, ambience, sound effects, and background music

AI Music
NEW

Generate original songs with custom lyrics and styles

CreationsCreations
PricingPricing
Prompt Library

Explore and discover high-quality AI prompts

Nano Banana Prompt Examples

Explore high-quality prompt examples for Nano Banana Pro

GPT Image 2 Prompt Examples
NEW

Browse GPT Image 2 prompts for social posts, posters, and product visuals

Seedance 2.0 Video Prompt Examples

Browse curated Seedance 2.0 video prompt examples with videos, authors, publish dates, and engagement metrics.

Prompt Toolkit

Prompt building tools and guides

AI Image Prompt Builder
NEW

Easily build complex prompts through a visual interface

Seedance 2 Prompt Guide

Dive deep into advanced features and best practices

API
NEW

Integrate Nano Banana Pro capabilities into your applications

Nano Banana Pro
AI ImageAI Video

Nano Banana Pro

Nano Banana Pro is the revolutionary AI-powered image generation and editing platform that delivers consistent character editing and scene preservation with superior performance.

Email: support@nanobananapro.site

  • Email

Product

  • Home
  • Features
  • Showcases
  • Ecommerce
  • Pricing

Other Services

  • AI Image Upscaler
  • Grok
  • Veo 3.1
  • Wan 2.5
  • Kling 4.0

Other

  • Blog
  • About Us
  • Climate Action Plan

© 2026 bananapro.site. All rights reserved.

  • Privacy Policy
  • Terms of Service
  • Refund Policy
  • Referral Program
  • Invoice Management
🇺🇸 English🇨🇳 中文🇰🇷 한국어🇯🇵 日本語🇪🇸 Español🇩🇪 Deutsch🇫🇷 Français🇷🇺 Русский🇸🇦 العربية🇧🇷 Português🇮🇹 Italiano
PricingCreate
Posters, product photos and GIFs in practiceExample prompts for posters, product photos and GIFsGPT Image 2.6 vs 2.5 on Nano Banana ProGPT Image 2.6 credits per imageGPT Image 2.6: practical questions

Images from text, edits from your own photos

GPT Image 2.6 on Nano Banana Pro supports posters, product scenes, illustrations and animated GIFs. A text prompt starts a new image; reference photos provide a starting point for edits. Flare, Sunburst and Economy share the same workspace, with different choices of aspect ratio, file format and resolution.

01

Three model versions

Flare · Sunburst · Economy

Flare suits quick iterations; Sunburst focuses on detail and consistent edits. Economy is the lower-credit 1K option, with a smaller selection of ratios and formats.

02

Transparent design assets

PNG · WebP

Transparent backgrounds at 1K suit stickers, icons and isolated product assets. PNG and WebP can preserve transparency when those formats are available for the selected version and mode.

03

Animated GIFs

GIF

GIF output is available in text-to-image mode. Frame count and time per frame can be automatic or set manually; the prompt describes the movement in the scene.

Posters, product photos and design assets

An event announcement, a shop listing and a presentation cover need different compositions and files. All three can start in the same workspace.

01

Event posters

Text-to-image can explore the main scene, headline placement and colours. Frequently changing dates and contact details can be added during later layout work.

02

Product scenes

An original product photograph provides the reference for edits to the background, light and colour palette.

03

Transparent assets

Isolated decorative elements suit web pages and presentations, with 1K background controls available for transparent output.

Generation settings and credits

The workspace brings together prompts, references and output settings. The Generate button displays the total for the selected settings and image count, with preview and original-file downloads in the results area.

01

Text and reference images

Text-to-image uses a written description for a new scene or GIF. Image editing accepts 1–16 references, with their roles and the details to retain described in the prompt.

02

Model and output settings

Flare, Sunburst and Economy offer different settings. Ratio, resolution and format are selected before generation; background controls are also available at 1K.

03

Animation settings and cost

Flare and Sunburst support 2–16 frames. Economy offers 4, 9, 16, 25, 36 or 64, or automatic selection. Frame duration can be automatic or set from 20 to 2,000 milliseconds in increments of 10. The Generate button displays the total credits.

04

Preview, download and recent history

The results area provides previews and original-file downloads. Workspace recent history restores a task’s settings for further edits to the prompt or output options.

Posters, product photos and GIFs in practice

A repair-event poster, a ceramic watering can and a compass animation cover three different jobs: lettering, reference editing and a repeating movement. The following examples explain the settings and image details involved in each.

Event posters

A bicycle wheel, a toolkit and a short event headline

01

Poster composition

A community repair poster can communicate its theme with a bicycle wheel, a toolkit and the headline “REPAIR WEEKEND”. A portrait canvas has room for the complete announcement; a small social preview relies more on a simple arrangement. In this example, the objects occupy the lower half and the title sits above them. That leaves a clear separation between the illustration and the words, without adding another subject to explain the event.

02

Headlines and event details

Short headlines can be included in the prompt, with quotation marks identifying the exact wording. Addresses, booking links and detailed schedules are often easier to add in a design application, where they remain convenient to update. The generated file is an image, so its lettering is part of the picture rather than an editable text layer. An empty panel in the composition can provide space for those changing details in the later layout.

03

The mobile preview

A phone-sized preview reveals whether the headline is too small or the main object disappears into the background. The full-size file is useful for checking spelling and smaller details. A landscape banner may need a different arrangement of the wheel and title, because cropping a portrait can remove the space reserved for copy. Fewer secondary objects can also make the event easier to recognise in a feed thumbnail, where the whole poster appears at once.

Product scenes

The same watering can in a different setting

01

The product photograph

A clear photograph of a watering can with a terracotta glaze can serve as the product reference for a garden scene. The prompt separates the original item from its new surroundings: the long spout, curved handle, glaze and proportions belong to the reference; the bench, wall and lighting belong to the new setting. This makes it easier to compare changes to the scene with changes to the object itself when the result appears in the preview.

02

Setting references

A complete handle and an unobstructed outline explain the structure more clearly than a crop or a heavy shadow. Another photograph can provide the setting, palette or arrangement, with that role stated in the prompt. Editing supports up to 16 references, though one product photo and one setting reference already cover many ordinary requests. Extra images are most useful when they supply a missing view or a specific detail, rather than repeating the same information.

03

Product details

Generated product scenes may include an extra opening, a thicker handle, a different spout angle or an invented decoration. Side-by-side comparison with the source makes these changes easier to spot. Materials, dimensions and included accessories still come from the real product information. A smaller follow-up edit can make a particular adjustment easier to track, especially when the surrounding scene already works and only one part of the product needs further attention.

Transparent assets

Separate assets for web pages and presentation covers

01

Subject and file format

A leaf cluster for a presentation cover only needs the leaves and some space around their outline. The prompt can describe their direction, viewing angle and approximate placement, while leaving out the vase, table or room. A transparent background is available at 1K, with PNG or WebP suitable for a static asset when those formats are offered. The simpler composition also leaves fewer surrounding objects to separate from the decorative element later.

02

Transparent edges

A pale fringe may be difficult to see on white and much clearer on a dark slide. Thin stems, gaps between leaves and soft shadows affect how an asset looks after placement. Genuine transparent areas show the background underneath; a checkerboard drawn into an opaque picture is only a pattern. A light and a dark test background provide a useful comparison, particularly for fine edges that are hard to judge in the original preview.

03

Asset size

A small corner decoration and a full-width image have different size requirements. Placement in the actual layout makes the scale and contrast easier to judge. Background controls are unavailable at 2K and 4K, and JPEG cannot store transparency. A larger file is not always a better fit for a small accent: the balance between the asset and the nearby text matters too, including how much empty space remains around a heading or illustration.

Website GIFs

A small needle movement in a fixed scene

01

Animation content

A compass beside a travel article can have a gently swaying needle while the circular rim and camera remain stationary. That is a shorter motion description than one combining camera rotation, scene changes and animated lettering. GIF output is available in text-to-image mode. Reference-based editing returns still images, so this workspace does not directly turn an uploaded photograph into an animation. The compass in this example is described in words as part of a new scene.

02

Frames and playback speed

Flare and Sunburst support 2–16 frames. Economy offers 4, 9, 16, 25, 36 or 64, or automatic selection. The time per frame can be automatic or set between 20 and 2,000 milliseconds in increments of 10. With the same frame count, a longer duration makes the animation slower. Four frames at 200 milliseconds each work out to a nominal 0.8-second sequence. The downloaded file provides the actual timing and playback, which is also what matters when the animation is placed in a publishing tool.

03

The loop transition

The transition back to the start makes a jumping needle, a changing rim or an unexpected camera movement especially noticeable. A small web animation also involves file size and the amount of attention it draws from nearby text. GIF output does not automatically optimise the file for a particular website. Playback in the destination application is another practical consideration, and a still version may suit a placement where continuous movement adds little to the page.

Common image problems

Composition, detail and playback involve different adjustments

01

Common image problems

Missing space beside a headline concerns composition; an incorrect handle concerns the reference and edit description; a pale fringe concerns the subject edge; a fast loop concerns frame duration. Each points to a different adjustment. Higher resolution or a request for “more detail” may leave the object in exactly the same position. A description of the visible issue gives the next generation a more specific change to work with than a general quality adjective.

02

Versions and settings

Flare is an option for trying compositions and visual directions, while Sunburst focuses on detail and reference editing. Economy offers 1K output with fewer ratios and formats at a lower credit cost. Results from a consistent prompt are easier to compare. When the prompt, version, ratio and background all change together, the contribution of any one setting becomes harder to judge. A brief record of those settings can help when several useful versions are being considered.

03

Credits and downloads

Resolution and batch size affect the total credits shown on the Generate button. Reopening a recent workspace task restores its original parameters, which can also be changed before another submission. Original-file downloads are available for later layout work or publication. Lettering, transparent edges and GIF playback still need a final review in the destination application, since a file that looks suitable in a preview may be used at a different size or against another background.

Example prompts for posters, product photos and GIFs

These examples cover text-to-image, reference editing and GIF output. Image 1 in the product example refers to an uploaded photograph of the item.

Community event poster

A wheel and toolkit establish the repair theme, with the headline above. Dates and contact details can be added in a design application.

A portrait poster for a community repair workshop. A bicycle wheel and a small toolkit occupy the lower half, with warm off-white paper and deep blue ink. The headline at the top reads “REPAIR WEEKEND”. There is no other lettering.

Watering can product scene

The reference provides the product’s appearance; the prompt provides its new setting and light.

The ceramic watering can from image 1 sits on a low wooden bench beside a sunlit wall. Its curved handle, long spout, terracotta glaze and proportions remain unchanged, with no extra openings or decoration.

Compass GIF

This example uses GIF output in text-to-image mode. Only the needle moves.

A flat illustration of a round compass on a pale background. The red needle sways slightly left and right, then returns to the centre. The rim and camera remain still, with no text or scene changes.

GPT Image 2.6 vs 2.5 on Nano Banana Pro

GPT Image 2.5 and 2.6 both support text-to-image generation and reference editing. This table covers the output formats, ratios and background options available on Nano Banana Pro.

SettingGPT Image 2.5GPT Image 2.6
Model versionsFlare and SunburstFlare, Sunburst and a lower-credit Economy option
Output choicesStill image generation and editingFlare and Sunburst can generate PNG, JPEG, WebP, animated GIF or editable SVG from text.
Resolution and ratio1K, 2K and 4K; 12 fixed ratios, with Auto depending on the generation modeFlare/Sunburst: 1K–4K, Auto plus 15 fixed ratios. Economy: 1K, Auto plus 7 fixed ratios.
Background controlsNo separate background selectorAuto, transparent or opaque at 1K; JPEG does not support transparency
Reference imagesUp to 16 images for an editUp to 16 images for an edit; GIF is text-to-image only
GPT Image 2.5 model page

GPT Image 2.6 credits per image

Prices below are for a square PNG at medium quality with no references, for accounts with a previous purchase. The price shown before generation reflects your settings, reference images and, for SVG, the authoring model.

Version1K (credits)2K (credits)4K (credits)
Flare122338
Sunburst122338
Economy7Not availableNot available

Accounts without a previous purchase use twice the credits shown. Each image in a batch is charged separately, and the button displays the total. Exact pixel dimensions depend on the selected aspect ratio and resolution tier.

Nano Banana Pro credit plans

GPT Image 2.6: practical questions

Answers about the settings available in this workspace, from output formats to credit costs.

Which version should I choose?

Flare suits quick iterations; Sunburst focuses on detail and consistent edits. Economy is the lower-credit 1K option, with a smaller selection of ratios and formats.

Can I generate a GIF from reference images?

GIF output is available in text-to-image mode. Reference-based edits produce still images: PNG, JPEG or WebP with Flare and Sunburst, and PNG with Economy.

How do transparent backgrounds work?

Background controls are available at 1K. PNG, WebP or GIF support a transparent background when available; JPEG cannot store transparency. Background controls are unavailable at 2K and 4K.

What happens when I reopen a task?

Workspace recent history restores a task’s prompt, references, version, ratio and output settings. Those settings can be adjusted for a new generation, charged at the current selection’s price.

Which aspect ratios and resolutions can I choose?

Flare and Sunburst offer Auto plus 15 fixed ratios at 1K, 2K or 4K. Economy offers Auto plus 7 fixed ratios at 1K. A resolution tier describes the output size class; the exact pixel dimensions also depend on the ratio.

How many reference images can I upload?

Image-to-image accepts 1–16 reference images. Their roles can be described by upload order in the prompt. Text-to-image, including GIF output, works without reference images.

Is GPT Image 2.6 free to use?

Generations use Nano Banana Pro credits. The table shows prices for accounts with a previous purchase; accounts without one pay double. The Generate button shows the total for your current settings and number of images before you submit.

Which file should I download for my project?

Flare and Sunburst can generate PNG, JPEG, WebP, animated GIF or editable SVG from text. Economy offers PNG and GIF. Download the original file to edit a vector in your design tool.

How do I edit with a mask or PNG annotations?

Choose Flare or Sunburst, switch to image editing and add a reference image. Open Advanced settings above the Generate button and upload either a mask or a transparent PNG with annotations. Use one of these options at a time. Image quality and composition complexity can also be adjusted there.

The GPT Image 2.6 image and GIF workspace

Text generation, reference editing and GIF output share one workspace. Credit costs depend on the version, resolution and number of images.

GPT Image 2.6 workspaceGPT Image 2.5 model page

GPT Image 2.6 — OpenAI's Next-Gen AI Image Model

Generate images, edit with references, or create GIFs with GPT Image 2.6.

Loading GPT Image 2.6...

Loading GPT Image 2.6...