Nano Banana 2.1 is Google's newest image generation and editing model. Google released it on October 6, 2026, and it is built on Gemini 3.6 Flash. In VidKiss it is now the default image model on Text to Image and Image to Image, and it has its own Nano Banana 2.1 model page with a tested example. VidKiss is an independent website and is not affiliated with Google.
1. What the model does
According to Google's announcement, Nano Banana 2.1 improves visual quality, prompt-guided editing, text rendering and subject consistency compared with earlier Nano Banana releases. Google says the model can take up to 14 reference images at once and keep up to four characters and ten objects consistent across them. It outputs at 1K, 2K or 4K.
Two of those numbers need a VidKiss-specific note. First, you can attach up to 8 reference images per generation in VidKiss; that is the upload limit for every image model on this website, so the model's higher limit is not available here. Second, the model is one model for both workflows: the same Nano Banana 2.1 handles text to image and image to image, so you do not switch models when you move from a fresh prompt to editing an upload.
2. What one image costs in VidKiss
Nano Banana 2.1 is priced per image by output resolution. As of October 8, 2026, one image costs 12 credits at 1K, 20 credits at 2K and 30 credits at 4K. Nano Banana 2 and Nano Banana Pro stay in the model menu at their existing prices, which are higher than Nano Banana 2.1 at every resolution. The quote beside Generate always shows the current price for the model and settings you picked, so treat that quote as the source of truth if this article and the workspace ever differ. See Pricing for credit packs.
3. Write a prompt the model can follow
Describe the subject, the setting, the lighting and the style, and say what must not appear. Short, concrete nouns work better than a pile of adjectives. If the image needs words in it, quote the exact text and ask for nothing else: for example, a label that reads exactly "WILDFLOWER HONEY" with "no other text, no logos". Better text rendering is one of the 2.1 improvements Google lists, but always check spelling in the result before you use it.
The launch poster on the model page is a real VidKiss result from October 8, 2026, made at the default 1K setting with a square 1:1 ratio. Its full prompt is published next to the image, so you can see exactly what produced the three-line 3D lettering. It is one result, not a promise that every run will look the same.
4. Edit or combine images
For image to image, upload the photo you want to change and describe the edit in one or two sentences: what to change, and what to keep. When you combine several references, say what each one contributes, for example which image sets the person, which sets the product and which sets the background. Keep the first request narrow, review the result, then change one thing at a time.
Upload only images you have the right to use. Uploaded images and generated results go through the same content checks as every other request on VidKiss.
5. Pick the resolution last
Start at 1K while you are still adjusting the prompt; it is the cheapest setting and the quickest to review. When the composition is right, run the same prompt at 2K or 4K for the final file. A new run at a higher resolution is a new generation, so details can change slightly between the 1K draft and the final image.
Finished images appear in the workspace and stay in your History for download. Ready to try it? Open Text to Image, check that Nano Banana 2.1 is selected, and read the quote before you generate.
