Gemini 3.1 Flash Image
Neural Network
Gemini 3.1 Flash Image is a Google neural network for generating and editing images based on text descriptions.
Max answer length
(in tokens)
Context size
(in tokens)
Image prompt
(per 1K tokens)
How it works Gemini 3.1 Flash Image?
Gemini 3.1 Flash Image, also known as Nano Banana 2, is a Google model for image generation and editing. It focuses on combining Pro-level visual quality with the speed of the Flash line: images are returned quickly while remaining detailed. You can work using both text descriptions and references — the final image can be refined precisely. In BotHub, the model is accessible without a VPN or foreign card, with payment in rubles via Russian cards and pay-as-you-go tokens that do not expire. Over 250 neural networks are available in one window: it's easy to compare options and switch models, and a unified OpenAI-compatible API eliminates the need to rewrite integrations. Data is transmitted with AES-GCM encryption and is not saved. For designers, the model saves hours on concepts and layout variations; for marketers, on creatives and product cards; for SMM specialists, on covers and post illustrations. Editors will find precise editing useful: changing backgrounds, removing unnecessary objects, or combining multiple references into one scene. Teams with corporate access have access to contracts, invoices, and an admin panel with limits.Frequently asked questions about Gemini 3.1 Flash Image
You can use generated results for commercial purposes. You own all rights to the content you create. The only restriction: make sure your prompt does not include copyrighted third-party material. You are responsible for respecting the rights to any input data.
nano-banana-2 from google-gemini creates images from text descriptions and processes existing images based on references. It is suitable for post visuals, banners, concepts, and illustrations, as well as for series of similar frames: batch mode helps prepare many variations in one run.
The model returns finished images, and specific output parameters depend on your request and generation settings. Our data indicates a context of 65,536 tokens and an output limit of up to 32,000 tokens — this is enough for a detailed prompt with a reference and detailed edits.
A separate reasoning mode for nano-banana-2 is not noted in our data — this is not a denial, we just haven't confirmed it. The model's main strength is image generation. If you need long logic and task analysis, switch to a text model in the same BotHub window.
Function calling and structured JSON are not noted in our data, so we won't promise them. If you are building a multi-model service, connect nano-banana-2 via the unified OpenAI-compatible BotHub API, and leave structured output to a text model — you won't have to rewrite the integration.
Images — yes: nano-banana-2 accepts an image as input and works in image-to-image mode, meaning it edits, completes, and varies your reference. Audio and video input are not noted in our data, so count on a combination of text and images.
We have no reliable data on prompt language, so we won't promise anything: the most honest way is to describe the task in the way that is convenient for you and compare the results. Access to the model in BotHub works without a VPN or foreign card, with payment in rubles.