Image Models

Choose an image model by workflow, speed, and the type of control you need.

Use the model that best matches the task rather than defaulting to one model for every image. The model menu and run estimate in xFigura are the source of truth for current availability, controls, and credit use.

Quick comparison: selected image models

This table covers six image models. Browse the other model pages in this section for additional options. Use these examples as starting points and review the result against your brief.

Model	Available modes in xFigura	Example uses	Start here when
GPT Image 2.5	Generate, edit	Flare for quick iteration; Sunburst for fine detail and precise edits	Your brief needs detailed visual work, with Sunburst as xFigura's default
Nano Banana 2 Lite	Generate, edit	Fast concepting and straightforward 1K edits	Turnaround and efficiency matter more than complex references
GPT Image 2	Generate, edit, inpaint	Instruction-heavy visuals, typography, layouts, precise edits	The image must follow detailed language or contain readable text
Nano Banana Pro	Generate, edit, inpaint	Complex compositions, reference-led work, brand and text consistency	Your brief combines references, text, and composition constraints
FLUX 2 Klein	Generate, edit	Rapid iteration and high-volume concepting	You need fast feedback before refining a direction
Seedream 4.5	Edit	Multi-reference edits, material changes, compositing	You are transforming or combining existing images

Choose by task

Fast concept exploration

Start with FLUX 2 Klein. Use it to test prompts, compositions, and broad visual directions quickly. Compare the results and refine the selected direction. Nano Banana 2 Lite, shown as Nano Banana Lite in xFigura, is another option for fast concepting and straightforward edits.

Fine detail and controlled revisions

Consider GPT Image 2.5 for detail-sensitive assets or edits with clearly defined constraints. Sunburst is xFigura's default variant; Flare is the speed-oriented variant. See the model page for their strengths and the difference between provider capabilities and xFigura controls.

Text, diagrams, or dense layouts

Start with GPT Image 2 or Nano Banana Pro. Provide the exact copy, hierarchy, placement, and any non-negotiable layout constraints in the prompt. Always review generated text before using the image externally.

Reference-driven generation

Start with Nano Banana Pro when the output must follow reference images, preserve a visual system, or combine multiple constraints into one coherent composition.

Editing and compositing

Start with Seedream 4.5 for multi-reference edits and material or object changes. Use GPT Image 2 when the edit must follow detailed instructions or use a mask for a tightly constrained region.

Targeted edits with a painted selection

Ideogram V4.5 Edit is available in the Image to Image node and for inpainting. Use it to edit a selected part of an existing image with a prompt. Precise Edit preserves the area outside your selection, and the quality control balances detail and credit use.

Nano Banana 2.1

Nano Banana 2.1 is Google's update to Nano Banana 2 for generation and conversational image editing, with improvements in text rendering, subject consistency, and visual quality. The model page covers provider capabilities and limitations. Check the xFigura model menu for current availability and supported controls.

About credits and timing

Credits and processing time can vary by model mode, resolution, quality, and the current model configuration. Review the estimate shown in xFigura before running a generation.

Continue

Open a model card above for inputs, strengths, limitations, and workflow guidance.