What is NanoBanana and how to use it (with Google AI Studio)
Aug 28, 2025
CompareAI.AI
NanoBanana is Google’s playful codename for its newest AI image model, officially Gemini 2.5 Flash Image. It’s built for high-quality image editing and image generation, with standout strengths like multi-step edits, character/identity consistency across edits, blending multiple photos, and targeted, natural-language transformations.
In this blog, you’ll learn everything about NanoBanana and the step-by-step instructions to edit images with Nono Banana AI.
TL;DR: If you want to try NanoBanana right now, use Google AI Studio (best, official, and consistent). You may also see it show up on LM Arena in head-to-head image-editing battles, but that route is inconsistent and not the preferred way to test it.
This is an official example of Gemini 2.5 Flash Image (Nano Banana) Image Editing.
What exactly is “NanoBanana”?
Codename & real name: “NanoBanana” is the internal nickname; the public model name is Gemini 2.5 Flash Image. Google’s own developer post explicitly introduces it “aka nano-banana.”
What it can do: Natural-language edits (e.g., “replace background with a beach”), multi-image fusion (combine two or more images convincingly), iterative multi-turn editing without breaking the subject’s identity, and style mixing. Google also applies SynthID watermarking to AI outputs for provenance.
Availability status: Labeled preview/trial in Google AI Studio as gemini-2.5-flash-image-preview.
The best way to use NanoBanana today: Google AI Studio (recommended)
Google AI Studio gives you a no-code, in-browser playground to create and edit images with NanoBanana. It’s the most direct and reliable way to try the model right now. Google AI Studio
Quick start (point-and-click)
Open Google AI Studio and start a new prompt. Make sure your project is set.
Switch the model to gemini-2.5-flash-image-preview (this is NanoBanana’s public name).
In the output settings, choose Image and text so the model returns images.
Insert media (upload a photo to edit) or start from text if you want to generate from scratch.
In the prompt box, describe your edit in plain language (e.g., “Replace the sky with golden hour clouds; keep the person’s face identical; add soft rim-light”). Click Prompt to run.
Iterate: Add follow-up instructions like “make the jacket matte black but keep the background unchanged.” Multi-turn editing is a core capability.
Tip: To blend images (e.g., put your pet into a vacation photo), upload both images and prompt: “Place the dog sitting on the car seat; match lighting; keep fur details consistent.”
Note on provenance: Images generated/edited by this model include an invisible SynthID watermark.
Step-by-step: Edit images with NanoBanana (hands-on guide)
Below is a practical flow you can follow inside Google AI Studio:
Start a new prompt in AI Studio → Switch model to gemini-2.5-flash-image-preview → set output to Image and text.
Upload your base image (Insert media). For best results, use a clear, well-lit photo.
Write a precise edit prompt. Examples you can paste:
“Remove the person in the background; keep lighting natural; do not change the subject’s face.”
“Change the T-shirt to deep navy; keep the logo untouched; add soft studio lighting; keep skin tone identical.”
“Replace the background with a misty pine forest; keep the model’s hair, face, and jacket identical.”
These natural-language, targeted edits are a core design goal of the model.
Refine with multi-turn edits. After the first result, add a follow-up like “reduce contrast by 10%, keep everything else unchanged.” Iterative editing is supported.
Blend two images (optional): Upload a second photo (e.g., your product cutout) and prompt “Place the product on the table in Image 1; match shadows and perspective.”
Export your result. Remember that outputs are watermarked via SynthID for transparency.
Check out our comparisons on image and video generative AI tools like Leonardo AI, Ideogram etc. with results and analysis over each AI tool, Or just judge them yourself by comparing the results of each ai tool, go to home now!.
What is NanoBanana and how to use it (with Google AI Studio)
Aug 28, 2025
CompareAI.AI
NanoBanana is Google’s playful codename for its newest AI image model, officially Gemini 2.5 Flash Image. It’s built for high-quality image editing and image generation, with standout strengths like multi-step edits, character/identity consistency across edits, blending multiple photos, and targeted, natural-language transformations.
In this blog, you’ll learn everything about NanoBanana and the step-by-step instructions to edit images with Nono Banana AI.
TL;DR: If you want to try NanoBanana right now, use Google AI Studio (best, official, and consistent). You may also see it show up on LM Arena in head-to-head image-editing battles, but that route is inconsistent and not the preferred way to test it.
This is an official example of Gemini 2.5 Flash Image (Nano Banana) Image Editing.
What exactly is “NanoBanana”?
Codename & real name: “NanoBanana” is the internal nickname; the public model name is Gemini 2.5 Flash Image. Google’s own developer post explicitly introduces it “aka nano-banana.”
What it can do: Natural-language edits (e.g., “replace background with a beach”), multi-image fusion (combine two or more images convincingly), iterative multi-turn editing without breaking the subject’s identity, and style mixing. Google also applies SynthID watermarking to AI outputs for provenance.
Availability status: Labeled preview/trial in Google AI Studio as gemini-2.5-flash-image-preview.
The best way to use NanoBanana today: Google AI Studio (recommended)
Google AI Studio gives you a no-code, in-browser playground to create and edit images with NanoBanana. It’s the most direct and reliable way to try the model right now. Google AI Studio
Quick start (point-and-click)
Open Google AI Studio and start a new prompt. Make sure your project is set.
Switch the model to gemini-2.5-flash-image-preview (this is NanoBanana’s public name).
In the output settings, choose Image and text so the model returns images.
Insert media (upload a photo to edit) or start from text if you want to generate from scratch.
In the prompt box, describe your edit in plain language (e.g., “Replace the sky with golden hour clouds; keep the person’s face identical; add soft rim-light”). Click Prompt to run.
Iterate: Add follow-up instructions like “make the jacket matte black but keep the background unchanged.” Multi-turn editing is a core capability.
Tip: To blend images (e.g., put your pet into a vacation photo), upload both images and prompt: “Place the dog sitting on the car seat; match lighting; keep fur details consistent.”
Note on provenance: Images generated/edited by this model include an invisible SynthID watermark.
Step-by-step: Edit images with NanoBanana (hands-on guide)
Below is a practical flow you can follow inside Google AI Studio:
Start a new prompt in AI Studio → Switch model to gemini-2.5-flash-image-preview → set output to Image and text.
Upload your base image (Insert media). For best results, use a clear, well-lit photo.
Write a precise edit prompt. Examples you can paste:
“Remove the person in the background; keep lighting natural; do not change the subject’s face.”
“Change the T-shirt to deep navy; keep the logo untouched; add soft studio lighting; keep skin tone identical.”
“Replace the background with a misty pine forest; keep the model’s hair, face, and jacket identical.”
These natural-language, targeted edits are a core design goal of the model.
Refine with multi-turn edits. After the first result, add a follow-up like “reduce contrast by 10%, keep everything else unchanged.” Iterative editing is supported.
Blend two images (optional): Upload a second photo (e.g., your product cutout) and prompt “Place the product on the table in Image 1; match shadows and perspective.”
Export your result. Remember that outputs are watermarked via SynthID for transparency.
Check out our comparisons on image and video generative AI tools like Leonardo AI, Ideogram etc. with results and analysis over each AI tool, Or just judge them yourself by comparing the results of each ai tool, go to home now!.
What is NanoBanana and how to use it (with Google AI Studio)
Aug 28, 2025
CompareAI.AI
NanoBanana is Google’s playful codename for its newest AI image model, officially Gemini 2.5 Flash Image. It’s built for high-quality image editing and image generation, with standout strengths like multi-step edits, character/identity consistency across edits, blending multiple photos, and targeted, natural-language transformations.
In this blog, you’ll learn everything about NanoBanana and the step-by-step instructions to edit images with Nono Banana AI.
TL;DR: If you want to try NanoBanana right now, use Google AI Studio (best, official, and consistent). You may also see it show up on LM Arena in head-to-head image-editing battles, but that route is inconsistent and not the preferred way to test it.
This is an official example of Gemini 2.5 Flash Image (Nano Banana) Image Editing.
What exactly is “NanoBanana”?
Codename & real name: “NanoBanana” is the internal nickname; the public model name is Gemini 2.5 Flash Image. Google’s own developer post explicitly introduces it “aka nano-banana.”
What it can do: Natural-language edits (e.g., “replace background with a beach”), multi-image fusion (combine two or more images convincingly), iterative multi-turn editing without breaking the subject’s identity, and style mixing. Google also applies SynthID watermarking to AI outputs for provenance.
Availability status: Labeled preview/trial in Google AI Studio as gemini-2.5-flash-image-preview.
The best way to use NanoBanana today: Google AI Studio (recommended)
Google AI Studio gives you a no-code, in-browser playground to create and edit images with NanoBanana. It’s the most direct and reliable way to try the model right now. Google AI Studio
Quick start (point-and-click)
Open Google AI Studio and start a new prompt. Make sure your project is set.
Switch the model to gemini-2.5-flash-image-preview (this is NanoBanana’s public name).
In the output settings, choose Image and text so the model returns images.
Insert media (upload a photo to edit) or start from text if you want to generate from scratch.
In the prompt box, describe your edit in plain language (e.g., “Replace the sky with golden hour clouds; keep the person’s face identical; add soft rim-light”). Click Prompt to run.
Iterate: Add follow-up instructions like “make the jacket matte black but keep the background unchanged.” Multi-turn editing is a core capability.
Tip: To blend images (e.g., put your pet into a vacation photo), upload both images and prompt: “Place the dog sitting on the car seat; match lighting; keep fur details consistent.”
Note on provenance: Images generated/edited by this model include an invisible SynthID watermark.
Step-by-step: Edit images with NanoBanana (hands-on guide)
Below is a practical flow you can follow inside Google AI Studio:
Start a new prompt in AI Studio → Switch model to gemini-2.5-flash-image-preview → set output to Image and text.
Upload your base image (Insert media). For best results, use a clear, well-lit photo.
Write a precise edit prompt. Examples you can paste:
“Remove the person in the background; keep lighting natural; do not change the subject’s face.”
“Change the T-shirt to deep navy; keep the logo untouched; add soft studio lighting; keep skin tone identical.”
“Replace the background with a misty pine forest; keep the model’s hair, face, and jacket identical.”
These natural-language, targeted edits are a core design goal of the model.
Refine with multi-turn edits. After the first result, add a follow-up like “reduce contrast by 10%, keep everything else unchanged.” Iterative editing is supported.
Blend two images (optional): Upload a second photo (e.g., your product cutout) and prompt “Place the product on the table in Image 1; match shadows and perspective.”
Export your result. Remember that outputs are watermarked via SynthID for transparency.
Check out our comparisons on image and video generative AI tools like Leonardo AI, Ideogram etc. with results and analysis over each AI tool, Or just judge them yourself by comparing the results of each ai tool, go to home now!.