Gemini 2.5 Flash Image, also known as nano-banana, is Google's advanced image generation and editing model introduced in August 2025. It supports blending multiple images into one, maintaining character consistency across edits, enabling targeted transformations through natural language, and leveraging Gemini’s world knowledge for enhanced image generation and editing.
Key Features
- Character Consistency: The model maintains the appearance of characters or objects across multiple prompts and edits, allowing placement in different environments or consistent brand asset creation.
- Prompt-Based Image Editing: Users can perform precise local edits such as blurring backgrounds, removing objects, changing poses, or adding color to black-and-white images using simple natural language prompts.
- Native World Knowledge: Gemini 2.5 Flash Image understands real-world context and semantics, enabling use cases like interactive educational tools that interpret hand-drawn diagrams and complex editing instructions.
- Multi-Image Fusion: The model can merge multiple input images into a single photorealistic image, such as placing objects into new scenes or restyling rooms with different textures or color schemes.
Accessibility and Pricing
The model is available via the Gemini API, Google AI Studio, and Vertex AI for enterprises. Pricing is set at $30 per 1 million output tokens, with each image typically costing about $0.039 based on 1290 output tokens per image. All other input and output modalities follow Gemini 2.5 Flash pricing.
Developer Tools and Support
Google AI Studio has been updated with a “build mode” that allows developers to quickly test, remix, and deploy AI-powered apps using Gemini 2.5 Flash Image. Template apps demonstrate features like character consistency, prompt-based editing, multi-image fusion, and interactive education tools. Developers can save code to GitHub or deploy directly from AI Studio.
Partnerships and Community
OpenRouter.ai has partnered with Google to bring Gemini 2.5 Flash Image to over 3 million developers, marking it as the first image-generating model on their platform. Additionally, fal.ai, a generative media developer platform, will help expand access to the broader developer community.
Image Authenticity
All images created or edited with Gemini 2.5 Flash Image include an invisible SynthID digital watermark to identify them as AI-generated or edited.
Example Usage (Python)
from google import genai
from PIL import Image
from io import BytesIO
client = genai.Client()
prompt = "Create a picture of my cat eating a nano-banana in a fancy restaurant under the gemini constellation"
image = Image.open('/path/to/image.png')
response = client.models.generate_content(
model="gemini-2.5-flash-image-preview",
contents=[prompt, image],
)
for part in response.candidates[0].content.parts:
if part.text is not None:
print(part.text)
elif part.inline_data is not None:
image = Image.open(BytesIO(part.inline_data.data))
image.save("generated_image.png")
Future Improvements
Google is actively working on enhancing long-form text rendering, improving character consistency reliability, and increasing factual accuracy in image details. Feedback is encouraged through the developer forum and social media channels.
Gemini 2.5 Flash Image is currently in preview with plans for a stable release soon, offering developers powerful new tools for creative and practical image generation and editing.
