Nano Banana 2
Google · AI imageNano Banana 2, also known as Gemini 3.1 Flash Image, is Google's latest AI image generation model. It combines Pro-level capabilities with Flash-style speed: higher resolution (up to 4K widescreen), better character consistency (up to five characters in one workflow), and accurate text in images. Google has positioned it as replacing both standard and Pro variants of earlier image models in many contexts.
VidMachine offers Nano Banana 2 so you can produce publication-ready visuals for thumbnails, start frames, and standalone images. When you need the best balance of quality and speed in the Google image lineup—without going to a separate, slower Pro-only pipeline—Nano Banana 2 is the default choice. It leverages mixture-of-experts and advanced quantization so that generation stays fast while quality reaches Pro levels.
The model can draw on Gemini's knowledge base and optional real-time search to create accurate infographics and detailed subject renderings. That makes it suitable for content that references real-world facts, places, or current events, as well as for creative and branded imagery.
Key features and benefits
Resolution and range
Nano Banana 2 supports output from 512 pixels up to 4K widescreen, so you can match the resolution you need for different platforms and print. Generation remains fast thanks to efficient architecture—mixture-of-experts activates only relevant pathways per prompt, and quantization reduces weight size without large quality loss. You get the flexibility to output small previews or full 4K assets from the same model.
Character consistency
The model can render up to five characters consistently within a single workflow, which helps with serial content, storyboards, and multi-panel visuals where the same people or characters appear. Identity and clothing can be maintained across images when you use consistent prompts or reference the same character descriptions. This is valuable for comic-style content, character-driven social series, and any workflow where visual continuity matters.
Accurate text in images
Nano Banana 2 generates legible, accurate text inside images, addressing a common limitation of earlier AI image models. That makes it suitable for infographics, captions, and branded graphics where words must be readable and correct. You can request specific phrases, labels, or signage and get coherent typography instead of garbled characters. This opens up use cases that were previously difficult or impossible with generic image models.
Object rendering
It can accurately render many different objects in one workflow (e.g. up to around 14), with good composition and detail for product shots and complex scenes. The model maintains spatial and semantic coherence so that multiple objects in a single image relate logically and look consistent in style and lighting. This supports product catalogs, scene composition, and detailed illustrations.
Efficiency and knowledge
The model uses mixture-of-experts and quantization for faster generation without sacrificing quality, and can draw on Gemini's knowledge and optional search for accurate, up-to-date subject matter. When your prompt involves real-world entities, facts, or current events, Nano Banana 2 can produce more accurate and relevant imagery than models that lack that grounding.
Technical specifications
Resolution512px up to 4K widescreen
CharactersUp to 5 consistent per workflow
ObjectsUp to ~14 per workflow
Official nameGemini 3.1 Flash Image
AccessGemini, Google AI Studio, Search, Google products
Use cases and applications
Nano Banana 2 is ideal for creators and brands who need high-quality, text-accurate images: thumbnails, start frames, infographics, product visuals, and social graphics. Use it when you want the best balance of quality and speed in the Google image lineup. Publication-ready assets, branded content with text, and character-consistent series all benefit.
Educators and businesses can use it for training materials, presentations, and marketing collateral where text and diagrams must be correct. E-commerce and product marketing benefit from the object-rendering and resolution range. On VidMachine, Nano Banana 2 is the go-to when you want to maximize image quality per credit while still getting fast turnaround.
Why this model
Nano Banana 2 replaces both standard and Pro variants of earlier Google image models in many contexts. On VidMachine it uses more credits per image than Nano Banana but delivers better consistency and text rendering. Choose it when quality and clarity matter most—hero thumbnails, key start frames, and any image that will be scrutinized at full resolution.
If you are on a tight credit budget or generating very large numbers of images, Nano Banana may be sufficient for many of them; reserve Nano Banana 2 for the images that drive clicks, conversions, or narrative.
How VidMachine uses it
Select Nano Banana 2 in your project's image model priority on VidMachine. It is used for generating images in AI Video (start frames) and AI Images projects. See Pricing and Docs for credit costs.
Each image generated with Nano Banana 2 consumes more credits than Nano Banana but typically less than Flux 2 Pro. Use it when you want the best Google image quality available on VidMachine.
What you should know
What resolution can Nano Banana 2 output?
Nano Banana 2 supports from 512 pixels up to 4K widescreen, depending on aspect ratio and use case.
Can Nano Banana 2 render text correctly?
Yes. It is designed for more accurate, legible text in images than earlier models.
How are Nano Banana 2 credits charged on VidMachine?
Each image generated with Nano Banana 2 uses a set number of credits. See Pricing for details.