Gemini 3.1 Flash Image
Google's top-ranked image generation and editing model, delivering world-class visual quality with strong text rendering, multi-subject consistency, and search-grounded creativity.
Google Text-to-image generation with search-grounded creativity
Gemini 3.1 Flash Image Preview, internally codenamed "Nano Banana 2," is Google DeepMind's Flash-tier image generation and editing model released in February 2026. It accepts image URL arrays alongside configurable input parameters, making it suitable for both generation and editing workflows. The model holds the number one ranking on the Artificial Analysis Image Arena and the Arena leaderboard for text-to-image generation, and every image it produces is invisibly watermarked via SynthID for provenance tracking.
The model is distinguished by its ability to maintain visual coherence across up to five characters and fourteen objects in a single image, its text rendering within generated images with multilingual support, and its use of real-time web search to inform image outputs. It supports flexible aspect ratios and upscaling up to 4K resolution. Gemini 3.1 Flash Image is available through the Gemini API, Google AI Studio, and Vertex AI, and is well suited for developers, designers, and businesses that need high-volume, high-quality image generation and editing at scale.
What Gemini 3.1 Flash Image supports
Text-to-Image Generation
Generates images from text prompts with support for flexible aspect ratios and upscaling up to 4K resolution.
Image Editing
Accepts image URL arrays as input to perform edits on existing images, enabling iterative visual workflows.
Multi-Subject Consistency
Maintains visual coherence across up to 5 characters and 14 objects within a single generated image.
Text Rendering in Images
Renders legible text within generated images with multilingual (i18n) support across multiple scripts and languages.
Search-Grounded Generation
Leverages real-time web search and knowledge to produce imagery that reflects accurate, up-to-date information.
SynthID Watermarking
Embeds an invisible SynthID watermark in every generated image to support authenticity verification and provenance tracking.
Configurable Output Settings
Exposes select-type input parameters for controlling generation options such as aspect ratio and output format.
Large Context Window
Supports a 131,072-token context window, allowing detailed prompts and multi-turn image generation sessions.
Ready to build with Gemini 3.1 Flash Image?
Get Started FreeCommon questions about Gemini 3.1 Flash Image
What is the context window for Gemini 3.1 Flash Image?
Gemini 3.1 Flash Image supports a context window of 131,072 tokens.
When was Gemini 3.1 Flash Image released and what is its training data cutoff?
The model was released in February 2026, and its training date is listed as February 2026.
What input types does Gemini 3.1 Flash Image accept?
The model accepts image URL arrays and two configurable select-type parameters, supporting both image generation from text and image editing workflows.
Where can developers access Gemini 3.1 Flash Image via API?
Developers can access the model through Google AI Studio, the Gemini API, and Vertex AI. Documentation is available on the Vertex AI docs site.
Does Gemini 3.1 Flash Image watermark its outputs?
Yes. Every image generated by the model includes an invisible SynthID watermark for authenticity and provenance tracking.
What is the maximum output resolution supported?
The model supports upscaling up to 4K resolution with flexible aspect ratio control.
What people think about Gemini 3.1 Flash Image
Community reception on Reddit has been largely positive, with users highlighting the model's top ranking on the Artificial Analysis Image Arena and noting its competitive pricing relative to the Pro-tier variant. Praise centers on text rendering quality, multi-subject consistency, and the value proposition of achieving high benchmark scores at a lower cost.
Some discussions focus on practical integration tips, including how to combine the model with other tools such as ChatGPT for hybrid workflows. Users have also noted the model's Flash-tier positioning as relevant for high-volume use cases where cost efficiency matters.
Google's Nano Banana 2 (Gemini 3.1 Flash Image Preview) takes #1 in Text to Image in the Artificial Analysis Image Arena at half the price of Nano Banana Pro!
Nano Banana 2 highlights and tips on combining it with ChatGPT
Parameters & options
If you want to edit an existing image, provide the URL(s) or variables
Explore similar models
Google Gemini 3 Pro Image
Google's most advanced image generation and editing model, built on Gemini 3 Pro, delivering professional-grade visual synthesis with multimodal reasoning.
Google Gemini 2.5 Flash Image
Google's state-of-the-art image generation and editing model that combines powerful creative control with Gemini's world knowledge.
Google Imagen 3
Google's Imagen 3 is a premium text-to-image model delivering photorealistic quality with exceptional text rendering precision and natural language understanding.
Black Forest Labs FLUX.2 [max]
Black Forest Labs' most powerful image generation and editing model, delivering professional-grade output with real-time web context and unmatched visual consistency.
Kling Kling Image O3
Kling AI's first image generation model, delivering high-quality text-to-image and image-to-image results with exceptional text rendering and up to 4K resolution support.
OpenAI GPT Image 1.5
OpenAI's flagship image generation model featuring precise editing, face preservation, and 4x faster generation speeds.
Start building with Gemini 3.1 Flash Image
No API keys required. Create AI-powered workflows with Gemini 3.1 Flash Image in minutes — free.