GPT Image 2.5 Image Generator

Create from a description or transform your reference images with GPT Image 2.5. Start with Flare for everyday creation, or switch to Sunburst when editing precision and fine detail matter most.

Loading editor…

From an idea to your next image

Choose Flare or Sunburst in the generator, then shape the result with your own instructions.

1

Describe or upload

Describe the subject, composition and style, or add reference images and explain what to change.

2

Choose your settings

Use Flare for speed or Sunburst for finer edits. Select the aspect ratio, resolution, quality and number of images.

3

Generate your images

Review the credit estimate before generating. Request 1–4 images, each processed as its own task.

4

Download or refine

Download the result or use it as a reference for another edit. State what should change and what should stay.

Two models, one creative workspace

Flare and Sunburst share the same controls and credit pricing on NovaImage. Choose the model that suits the task.

GPT Image 2.5 Flare

GPT Image 2.5 Flare

The speed-focused option for everyday image creation, exploring ideas and trying visual variations. Flare is selected when you first open this page.

GPT Image 2.5 Sunburst

GPT Image 2.5 Sunburst

The detail-focused option for precise reference edits and carefully directed compositions. Allow more generation time than with Flare.

Create with up to 16 references

Create with up to 16 references

Combine reference images with written instructions. Specify each image’s role and describe the details you want to preserve or change.

Control format and quality

Control format and quality

Choose from 15 aspect ratios, 1K / 2K / 4K and low / medium / high quality. The displayed resolution is a size tier; exact dimensions depend on the aspect ratio.

Make images for your next project

Start with a clear brief and choose the model for your workflow.

Product scenes

Product scenes

Place a product reference in a new setting and describe the background, lighting and composition.

Posters and ads

Posters and ads

Explore campaign visuals. Provide the exact headline and desired placement, then review text before publishing.

Social content

Social content

Create portrait, square or landscape images for different placements using a shared visual brief.

Reference edits

Reference edits

Request changes to clothing, objects or a scene, and explicitly identify the elements to preserve.

GPT Image 2.5 questions

Both are GPT Image 2.5 models for image generation and reference editing. Flare prioritizes speed; Sunburst focuses on precision and fine detail and takes longer. This page defaults to Flare, and you can switch models in the generator.

Yes. Add up to 16 JPEG, PNG or WebP reference images, each no larger than 10MB in the uploader, and describe the changes. Explain what each reference represents and what should remain unchanged.

Both models offer 15 aspect ratios, 1K / 2K / 4K resolution tiers and low / medium / high quality. Initial settings are 1:1, 1K, low and one image. This page does not offer transparent backgrounds, masks or custom pixel dimensions.

Choose 1–4 images. NovaImage submits a separate task for each image, so results may finish at different times. Confirmed failed tasks are refunded individually; successful images are retained.

Flare and Sunburst use the same pricing as GPT Image 2 on NovaImage. At low quality and 1K, one text-to-image result costs 4 credits and one reference edit costs 6. Other settings cost more; review the total shown in the generator.

Yes. This page saves your model choice, settings and reference images in a local draft for up to 24 hours. Return using the same browser. An explicit model link can start with a different model; clearing browser storage removes local drafts.

Create with GPT Image 2.5

Start with Flare, or choose Sunburst for your next detailed edit.

Start creating

NovaImage is an independent video and image creation platform using third-party AI models. Learn more