Create images or edit references online with GPT Image 2.6. Sign in to use account credits: 1K 24, 2K 48, or 4K 80, plus 7 for each reference image.
GPT Image 2.6 is OpenAI’s image generation and editing release with two API variants: Flare and Sunburst. OpenAI says the release generates images 50% faster than before, while Flare improves layouts and Sunburst focuses on higher-quality output and consistency across edits.
On EaseGen, both variants create images from text or edit up to 16 reference images, with 1K, 2K, and 4K output. Choose Flare to explore layouts and ideas, or Sunburst for detailed images and repeated edits.
See what changed in GPT Image 2.6 and which capabilities are available in EaseGen today.
OpenAI says Images 2.6 generates images 50% faster than before and that Flare improves layouts for posters and product visuals. On EaseGen, choose Flare to explore compositions and variations. Actual generation time depends on the request and service conditions.
OpenAI says Sunburst keeps image details more consistent across edits at medium or high quality. On EaseGen, upload an image, describe the revision and name the product, character or design details to preserve. Review each result for unintended changes; EaseGen does not currently offer a separate quality selector.
GPT Image 2.6 brings image markup to the OpenAI API: a circle, sketch or note can identify where an edit should happen. This can make a local change easier to describe. EaseGen currently supports text instructions and reference images, but does not offer a markup tool.
GPT Image 2.6 can create a short animated GIF from a prompt or a still image, such as a mascot waving or steam rising from a cup. GIF output is available through the model API; the current EaseGen generator produces still images only.
GPT Image 2.6 can create SVG graphics from a text prompt. SVGs scale without losing sharpness, making them useful for simple icons, badges and web graphics. SVG output is available through the model API but is not currently offered in EaseGen.
The new complexity setting controls how much planning goes into a request. OpenAI recommends low complexity for many quick edits and suggests trying medium or high for scenes with multiple elements or constraints. EaseGen does not currently expose this setting.
Explore the original prompts and results, then adapt them to your own ideas.
Create a landscape 3:2 poster for a fictional botanical exhibition. Show a close-up of a vivid red poppy and deep green leaves against a warm white background. Large, clearly legible black headline reads exactly "BOTANICAL FORMS". Smaller text reads exactly "SPRING EXHIBITION". Arrange the type and flower with generous spacing, crisp editorial typography and natural photographic detail. No real brands, logos or watermark.
Create a landscape 4:3 studio product photograph of a single unbranded cobalt blue glass perfume bottle with a brushed silver cap, standing on a pale gray stone pedestal. The entire bottle is visible with generous margins. Soft directional daylight casts a distinct shadow; show realistic reflections and glass thickness. Add a small orange flower beside the pedestal. Clean premium composition, no text, logos or watermark.
Edit the reference product image. Keep the exact same cobalt blue glass bottle, silver cap, shape, scale and camera angle. Replace the gray stone pedestal and studio background with a sunlit green botanical setting, placing the bottle on a moss-covered stone. Preserve realistic glass reflections and show the entire bottle. Remove the orange flower. No text, logos or watermark.
The product photograph above was used as the reference for this edit.
Both variants use medium quality and the same EaseGen inputs. The table shows base credits without references; add 7 credits per reference image. Choose by the kind of work you need to finish.
| Variant | Best Fit | References | Resolution | Credits by Resolution |
|---|---|---|---|---|
| GPT Image 2.6 Flare | Fast drafts, layouts, posters, product visuals, and quick variations | Up to 16 | 1K / 2K / 4K | 1K 24 · 2K 48 · 4K 80 |
| GPT Image 2.6 Sunburst | Detailed visuals, reference edits, and multi-round refinement | Up to 16 | 1K / 2K / 4K | 1K 24 · 2K 48 · 4K 80 |
Current EaseGen settings. For 2.6 and 2.5, the table uses Flare; Sunburst is also available.
| On EaseGen | GPT Image 2.6 Flare | GPT Image 2.5 Flare | GPT Image 2 |
|---|---|---|---|
| Create and edit | Text and references | Text and references | Text and references |
| Reference images | Up to 16 | Up to 16 | Up to 16 |
| Resolution | 1K / 2K / 4K | 1K / 2K / 4K | 1K / 2K / 4K |
| Aspect ratios | 8 + Auto | 8 + Auto | 15 + Auto |
| Choose for | Newer Flare and Sunburst workflows | An established 2.5 workflow | A wider choice of aspect ratios |
Compare the settings currently available on EaseGen; output quality depends on the prompt and task.
| On EaseGen | GPT Image 2.6 Flare | Nano Banana 2.1 | Seedream 5.0 Lite |
|---|---|---|---|
| Create and edit | Text and references | Text and references | Text and references |
| Reference images | Up to 16 | Up to 10 | Up to 14 |
| Resolution | 1K / 2K / 4K | 1K / 2K / 4K | 1K / 2K / 4K |
| Aspect ratios | 8 + Auto | 14 + Auto | 8 |
| Choose for | OpenAI Flare or Sunburst workflows | More aspect ratio choices | Everyday image creation and editing |
Step 1
Start with Flare for quick concepts and layout exploration. Choose Sunburst when the image needs detailed control or several editing rounds.
Step 2
Use Text to Image for a new scene, or Image to Image with one or more references. Describe what should change and what must stay consistent.
Step 3
Choose a supported ratio and 1K, 2K, or 4K output. Both variants use medium quality: base costs are 24, 48, and 80 credits respectively, plus 7 per reference image. Sign in and review the total before generating.
Create an AI video from an image. Add a product photo or character reference, then describe the movement you want. Image to Video
Turn your ideas into images and videos in one creative workspace.
