Google's native Gemini 3 image models reason about a prompt before drawing it. Gemini 3.1 Flash at 1 credit, Gemini 3 Pro at 2 credits, text to image and editing, legible text, five aspect ratios, no watermark. Available on every Aura AI plan.
Gemini 3 image generation is Google's name for the native image capability of the Gemini 3 model family. Two models carry it. Gemini 3 Pro Image, released on November 20, 2025 and built on Gemini 3 Pro, is the high-fidelity tier: it plans a composition with the reasoning of the language model, renders long and multilingual text legibly, blends up to 14 reference images, keeps up to five subjects consistent, outputs at 1K, 2K or 4K and can ground a picture in Google Search results. Gemini 3.1 Flash Image, released on February 26, 2026 and built on Gemini 3 Flash, is the high-efficiency counterpart: faster and cheaper, with optional thinking, output from 0.5K up to 4K, improved international text rendering and conversational editing.
You may know both under their nicknames. Google calls Gemini 3 Pro Image "Nano Banana Pro" and Gemini 3.1 Flash Image "Nano Banana 2", continuing the name that the original Gemini 2.5 Flash Image made popular. On Aura AI the two brandings live side by side: the Nano Banana entries are the nickname-branded options at 3 credits, while the Gemini 3 Pro and Gemini 3.1 Flash entries described on this page are the Gemini-branded image endpoints at a lower credit cost of 2 and 1.
On Aura AI, Gemini 3.1 Flash (1 credit) and Gemini 3 Pro (2 credits) are available in the Text to Image tool in five aspect ratios, and as Gemini 3.1 Flash Edit (1 credit) and Gemini 3 Pro Edit (2 credits) in the image editor. Output is delivered without watermark. They sit next to Imagen 4, Google's other image line, and to MAI Image 2.5 and FLUX 2 in the same selector.
Six things the Gemini 3 image models do differently
Both models are native image outputs of a Gemini language model. Gemini 3 Pro Image uses reasoning-enhanced composition and Gemini 3.1 Flash Image has a thinking mode, so complex briefs with several constraints are planned before pixels are produced.
Long passages, headlines and labels come out readable, and 3.1 Flash adds improved international text rendering. Infographics, posters and multilingual ads are the natural fit.
Upstream, Gemini 3 Pro Image blends as many as 14 input images and keeps the identity of up to five people consistent. Group shots, product ranges and character casts stay recognisable.
Google lists 1K, 2K and 4K output for Gemini 3 Pro Image and 0.5K to 4K for Gemini 3.1 Flash Image. Large prints and full-width web heroes do not need a separate upscaling step upstream.
Both models can pull current facts from Google Search when generating, which is how an infographic about a real event or a diagram of a real product can be accurate rather than invented.
Describe a change in plain language and the model applies it to your image. On Aura AI this runs as Gemini 3.1 Flash Edit at 1 credit or Gemini 3 Pro Edit at 2 credits in the image editor.
Four steps from prompt to a finished image
Flat cost per image, identical for every aspect ratio
| Variant | Mode | Credits per image | Best for |
|---|---|---|---|
| Gemini 3.1 Flash | Text to image | 1 credit | Drafts, volume, social visuals |
| Gemini 3 Pro | Text to image | 2 credits | Text-heavy layouts, multi-subject scenes, final renders |
| Gemini 3.1 Flash Edit | Image edit | 1 credit | Quick conversational changes |
| Gemini 3 Pro Edit | Image edit | 2 credits | Precise edits, identity preservation |
How the Gemini-branded endpoints compare to their siblings on Aura AI
| Feature | Gemini 3 Image | Nano Banana | Imagen 4 | MAI Image 2.5 |
|---|---|---|---|---|
| Maker | Microsoft AI | |||
| Underlying model | Gemini 3.1 Flash Image, Gemini 3 Pro Image | Same models, nickname branding | Imagen 4 | MAI-Image-2.5 |
| Text to image credits | 1 (Flash), 2 (Pro) | 3 | 1 | 2 |
| Image edit credits | 1 (Flash), 2 (Pro) | 2 (Edit), 3 (Nano Banana 2 Edit) | Not on Aura AI | Not on Aura AI |
| Reasoning or thinking mode | ✓ Yes | ✓ Yes | No | Not published |
| Reference images (upstream) | ✓ Up to 14 (Pro) | ✓ Up to 14 (Pro) | Text only | Editing not on Aura AI |
| Max output upstream | 4K | 4K | 2K | Not published |
| Plan | All plans | All plans | All plans | All plans |
Where reasoning, text rendering and references pay off
Charts, step-by-step explainers and labelled diagrams with every word readable. Gemini 3 Pro plans the layout first, which is why the text lands where it should.
One creative, several languages. The improved international text rendering in Gemini 3.1 Flash makes localised social ads a 1 credit job per variant.
Keep several characters recognisable across a comic, a storyboard or a campaign. Gemini 3 Pro preserves identity for up to five subjects across scenes upstream.
At 1 credit per image, Gemini 3.1 Flash is the model for exploring twenty thumbnail concepts before spending 2 credits on the final with Gemini 3 Pro.
Common questions about Gemini 3 image generation on Aura AI
Google's reasoning image models from 1 credit per image, text to image and editing, no watermark. Available on every plan, worldwide.