Nano Banana 2

Image GenerationImage EditingUp to 4K

Nano Banana 2 on APIXO maps to Google’s stable Gemini 3.1 Flash Image model, the high-efficiency counterpart to Gemini 3 Pro Image. It supports prompt-led creation and natural-language editing with up to 14 image inputs, improved multilingual text rendering, stronger visual consistency, 1K–4K output, and extreme panoramic framing for high-volume image workflows.

Upstream Model

Gemini 3.1 Flash Image

APIXO Price

$0.05–$0.12 / Image

Resolution

1K / 2K / 4K

Image Inputs

Up to 14

Released

May 28, 2026

Loading workspace...

Capabilities

More control over detailed prompts, references, and layouts

Multi-Constraint Prompts

Follow layered directions covering subjects, actions, lighting, spatial relationships, camera choices, and layout requirements more reliably than earlier Flash image generation.

Natural-Language Editing

Add, remove, restore, replace, or restyle visual elements through written instructions. APIXO exposes no mask field, so the prompt must clearly identify the intended edit.

Multi-Image Composition

Combine several supplied images as multimodal context for composite scenes, coordinated assets, product arrangements, or subject-led creative concepts.

Subject and Object Consistency

Carry recognizable subjects and objects across changed environments and compositions with improved consistency, without guaranteeing exact identity or geometry.

Multilingual Text Rendering

Produce more reliable embedded text and localized visuals for posters, advertisements, menus, editorial layouts, and interface concepts across multiple languages.

World-Aware Generation

Apply broad world knowledge and spatial reasoning when depicting places, objects, processes, and context-rich scenes. Factual details still require human verification.

APIXO Workflows

Create new imagery or transform supplied visual material

Text-to-Image Generation

Submit a prompt of up to 20,000 characters, then select an APIXO resolution, aspect ratio, and output format. Use this workflow for advertisements, editorial graphics, concept art, product scenes, illustrations, and information-rich layouts without source images.

Create from text

Image-to-Image Editing

Provide one to 14 public, directly accessible image URLs with a required instruction. APIXO supplies these files as editable inputs or general multimodal context rather than separate character-, object-, or style-reference systems. Each request is independent.

Edit with images

APIXO Configuration

Verified output controls

The production options exposed on APIXO’s Nano Banana 2 route.

1–20,000 Characters

Prompt Length

15 Options

Aspect Ratios

1:8 to 8:1

Framing Range

PNG / JPEG / JPG

Output Formats

Async / Callback

Delivery

15 Days

Result URL Retention

Production Uses

Build adaptable assets for campaigns and creative systems

Global Marketing

Localize Complete Advertisements

Translate embedded copy while adapting surrounding imagery for regional audiences. Improved multilingual rendering supports posters and promotional graphics, but teams should verify spelling, line breaks, legal text, logos, and cultural details before publication.

Nano Banana 2 Recomposed Product Photography

Product Content

Recompose Product Photography

Combine product images, environmental references, and written art direction to explore new campaign scenes. Select higher-resolution output for final candidates while checking packaging, proportions, materials, labels, and geometry against approved source assets.

Display Creative

Design Extreme-Wide Assets

Generate conventional layouts alongside 1:4, 4:1, 1:8, and 8:1 compositions for banners, panoramic headers, installations, and specialized display placements without manually extending a standard canvas.

Creative Automation

Produce Structured Variations

Combine detailed prompts and visual inputs to generate campaign variants, story concepts, educational graphics, or editorial assets. Use 1K for economical iteration, then move selected directions to 2K or 4K.

Notes & FAQ

Integration boundaries for Nano Banana 2

Implementation Notes

APIXO’s route represents Gemini 3.1 Flash Image, not the original Gemini 2.5 Flash Image, Gemini 3 Pro Image, or Nano Banana 2 Lite.

APIXO exposes 1K, 2K, and 4K resolution; Google’s upstream 0.5K option is not available on this route.

APIXO’s public materials currently describe Google Search inconsistently: the model page fixes google_search to false, while the detailed API reference lists it as an optional boolean. Treat Search Grounding as unavailable until the live schema confirms that true is accepted.

APIXO exposes no configurable Thinking, seed, mask, transparency, safety-level, or output-count field for this route.

APIXO supports asynchronous polling or webhook delivery; completed result URLs remain available for 15 days by default.

Google embeds an invisible SynthID watermark in generated images; this is distinct from a visible watermark.

Frequently Asked Questions

No. Nano Banana 2 is Gemini 3.1 Flash Image, Google’s versatile balance of intelligence, speed, quality, and cost. Nano Banana Pro is the separate Gemini 3 Pro Image model for more demanding professional asset production.

Not as a confirmed active control. APIXO’s public model page currently fixes google_search to false, while the detailed reference describes it as an optional field. Verify the live schema before enabling or advertising grounded generation.

APIXO charges per generated image: $0.05 at 1K, $0.08 at 2K, and $0.12 at 4K. Its public billing documentation lists no additional charge based on mode, reference count, aspect ratio, or output format.

No persistent multi-turn context is documented. Treat every generation as a one-shot task and include the complete prompt and all required source images again when continuing an earlier edit.

Review identity, hands, anatomy, embedded text, translations, factual details, logos, packaging, product geometry, and reference fidelity. Safety filtering can reject content, and repeated requests are not guaranteed to reproduce identical images.