Gemini 3 Pro Image Preview
Neural Network
Gemini 3 Pro Image Preview (Nano Banana Pro) is Google's model for image generation and editing.
Max answer length
(in tokens)
Context size
(in tokens)
Image prompt
(per 1K tokens)
How it works Gemini 3 Pro Image Preview?
Nano Banana Pro is Google's solution for image generation and editing, built on Gemini 3 Pro. Compared to the first version of Nano Banana, multimodal reasoning and real-world knowledge are enhanced, allowing the model to better understand the task's intent rather than just the formal frame description. It works for both text-to-image generation and editing existing images via reference. In BotHub, the model is accessible from Russia without a VPN or foreign card: you pay with a Russian card in rubles only for used tokens, with no subscription. Over 250 models are available in one window, making it easy to compare results. For integrations, there is a unified OpenAI-compatible API and corporate terms with contracts, invoices, and an admin panel. Designers create concepts and visuals, marketers build campaign creatives and banners, editorial teams generate illustrations, product teams make screen mockups, and SMM specialists refine existing shots without new photoshoots.Frequently asked questions about Gemini 3 Pro Image Preview
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.
This is a Google model for image generation: it creates images from text descriptions and processes uploaded references. It is suitable for illustrations, concepts, banners, covers, social media visuals, and product cards, as well as serial generation of variants via batch mode and the unified BotHub API.
You get a finished raster image as output, with exact dimensions and aspect ratios set by request parameters. A 65,536-token context allows for detailed descriptions alongside references, with up to 32,000 tokens allocated for the response. The model handles prompt details accurately.
A separate reasoning mode is not noted in our data, so we neither confirm nor deny it. Technically, it is a Gemini-line model with text input and image output, so it carefully parses complex, multi-part descriptions. Test it with your own prompt: one generation will say more than any promises.
Function calling and structured JSON output are not noted in our data, so we cannot confirm them: the model's profile is images. However, requests go through the unified OpenAI-compatible API of BotHub, and you can easily connect a text model for JSON alongside it without rewriting the integration.
Images — yes: the model works in image-to-image mode, so you can attach a reference and ask to rework it, change the style, create variations, or complement the composition. There are no notes about audio or video input in our data, so focus on the combination of text and images.
The prompt language is not described in our data, so we won't make any promises in advance — it's easier to check with a single request. On the BotHub side, Russian is available everywhere: interface, Telegram bot, payment with Russian cards without VPN or foreign cards. If it doesn't suit you, switch to another model in the same window.