GPT Image 2.5 Sunburst
An OpenAI image model for reference-led editing, product transformations, structured graphics, and 4K output.
Key facts
| Developer | OpenAI |
|---|---|
| Official model ID | gpt-image-2.5-sunburst |
| Released | 2026-09-08 |
| Availability on Banana Pie | Available now |
| Credits from | 30 credits / image |
| Reference images | Up to 8 |
OpenAI positions GPT Image 2.5 Sunburst as the precision-editing model of the GPT Image 2.5 pair: it trades speed for control and changes only what the prompt asks to change (OpenAI announcement, 2026-09-08). In Banana Pie's evaluation it is positioned for reference-heavy production work where objects, garments, text, layouts, or products must carry into a new image. Its 8-reference allowance and 4K ceiling are practical advantages, but the available evidence does not establish repeatability or superiority over sibling models.
Strengths
- Reference-heavy workflows
Accepts up to 8 reference inputs, with published selections covering Six-reference object binding and edit as well as identity and garment transfer.
- Editing-oriented coverage
The fixed suite includes in-image text replacement, coherent scene relighting, cross-ratio outpainting, and product preservation across scene change.
- Commercial asset scenarios
Advertising composition, multilingual typography, and structured UI graphic scenarios make the model relevant to asset-production workflows.
- 4K output
The listed maximum resolution is 4K, supporting workflows that need larger final images.
Limitations
- Repeatability is unproven
Every scenario records only run 1, so the evidence does not establish consistency across repeated generations.
- No comparative ranking
The published evidence contains selected outputs but no head-to-head scores against sibling models.
- Variable observed runtime
Published selections ranged from 69.0s to 154.8s, so creators should not assume a fixed turnaround.
- Iteration cost
Each generation starts at 30 credits, creating a meaningful cost floor for rapid prompt exploration.
Choose GPT Image 2.5 Sunburst when
- You need to combine several source images, with support for up to 8 references.
- Your task involves product preservation, garment transfer, relighting, outpainting, or text replacement.
- You are producing advertising layouts, multilingual typography, or structured UI graphics.
- You need output up to 4K and can budget at least 30 credits per generation.
Consider an alternative when
- Your credit budget cannot accommodate the 30-credit minimum.
- You need a tightly predictable turnaround rather than the observed 69.0s to 154.8s range.
- You require evidence of repeatable results across reruns.
- You need a proven head-to-head winner against another Banana Pie model.
Depending on the task, these can be a better fit: GPT Image 2.5 Flare · Nano Banana Pro
Real-world results
Same fixed prompts we run on every model - one published result per model, with prompts and settings you can inspect.
Prompt
A premium product photograph of exactly one transparent rectangular perfume bottle, half filled with amber liquid, standing upright on wet black stone. Light comes from the upper left, creating coherent refraction, a contact shadow, and one subtle reflection. No text, logo, plants, or extra objects.
Settings
- Resolution: 1K
- Aspect Ratio: 1:1
Prompt
Design a clean cream-colored vertical event poster. Show exactly four centered text lines and no other text: "MOONLIGHT MARKET", "月光市集", "18 OCT", and "RIVER HALL". Render "月光市集" in red and all other lines in black.
Settings
- Resolution: 1K
- Aspect Ratio: 2:3
Prompt
A photorealistic close-up of an adult potter shaping a clay bowl on a spinning wheel. Both hands are fully visible, each with five natural fingers touching the clay. No other people or hands. Soft window light.
Settings
- Resolution: 1K
- Aspect Ratio: 3:4
Prompt
On a matte gray table, exactly three red wooden cubes form a row on the left, exactly two blue glass spheres sit on the right, and one yellow ceramic mug stands centered behind them. The mug handle points right. No other objects or text.
Settings
- Resolution: 1K
- Aspect Ratio: 4:3
Prompt
Create a vertical social ad for a fictional running shoe named AERO. Keep the top 15 percent empty. Directly below it, place the headline "RUN LIGHT" in the upper-left. Show exactly one silver shoe in the lower-right, an orange trail curving from the bottom-left, and a round badge reading "42 KM". No other shoes or text.
Settings
- Resolution: 1K
- Aspect Ratio: 9:16
Prompt
Create a clean horizontal pricing comparison graphic titled "CHOOSE YOUR PLAN". Use exactly three equal columns labeled "STARTER", "PRO", and "TEAM". Under each column, show exactly three aligned rows labeled "PROJECTS", "STORAGE", and "SUPPORT", followed by one blue button labeled "SELECT". White background, dark navy text, no additional columns or text.
Settings
- Resolution: 1K
- Aspect Ratio: 16:9
Reference images
Prompt
Use Reference 1 as the base tabletop scene. Replace the red mug with the exact dark navy-blue mug from Reference 2 in the same position and orientation. Replace the folded white towel with the exact green-and-white striped napkin from Reference 3 in the same folded area. Replace the green pear with exactly one orange from Reference 4 in the same position. Replace the blue notebook with the exact mustard-yellow hardcover notebook from Reference 5 in the same position. Add exactly the two lemons from Reference 6 to the right of the mug. Preserve the wooden table, camera, framing, wood grain, lighting, shadows, and all other spatial relationships from Reference 1. Do not copy the white product backgrounds from References 2–6, duplicate any asset, or add other objects.
Settings
- Resolution: 1K
- Aspect Ratio: 4:3
Reference images
Prompt
Replace only the sign text with exactly "NIGHT OWL". Preserve the original font style, spacing, perspective, sign material, lighting, and everything else.
Settings
- Resolution: 1K
- Aspect Ratio: 3:2
Reference images
Prompt
Use the portrait in Reference 1 for the person and the isolated jacket in Reference 2 for the garment. Dress the person from Reference 1 in the exact jacket shown in Reference 2. Preserve the person's identity, face, expression, skin, hair, hands, pose, body proportions, background, framing, and lighting from Reference 1. Preserve the jacket's material, color, collar, buttons, pockets, and sleeve patch from Reference 2. Do not copy the ghost mannequin or white product background.
Settings
- Resolution: 1K
- Aspect Ratio: 3:4
Reference images
Prompt
Change the lighting to warm golden-hour sunlight entering from the left window. Do not move, add, remove, or redesign any object. Update highlights and shadows coherently.
Settings
- Resolution: 1K
- Aspect Ratio: 16:9
Reference images
Prompt
Expand the canvas to a 16:9 landscape by naturally continuing the beach, ocean, and sky on both sides. Keep the complete original image centered without cropping, stretching, letterboxing, or modifying it.
Settings
- Resolution: 1K
- Aspect Ratio: 16:9
Reference images
Prompt
Place the sneaker on a wet outdoor basketball court at dusk. Preserve the exact sneaker shape, black geometric side mark, materials, stitching, sole geometry, and camera angle. Add physically coherent contact, reflections, and dusk lighting.
Settings
- Resolution: 1K
- Aspect Ratio: 1:1
Credits & pricing
The minimum listed cost is 30 credits, so budget at least 30 credits for each generation. Higher-cost settings or a complete pricing table are not included in the available facts.
Frequently asked questions
Is GPT Image 2.5 Sunburst available now?
Yes. It is listed as available in Banana Pie.
How many reference images can I use?
It accepts up to 8 reference inputs. The published suite includes Six-reference object binding and edit, plus multi-reference identity and garment transfer.
What is the maximum output resolution?
The listed maximum resolution is 4K.
How many credits does it cost?
The minimum listed cost is 30 credits per generation. The available facts do not specify a complete price table for other settings.
How long does generation take?
Published selections ranged from 69.0s to 154.8s across the fixed prompt suite. These are observed run times, not a guaranteed turnaround.
How we test & disclosure
Conclusions come from a fixed prompt suite and published auto:only-run selections covering product work, typography, anatomy, structured graphics, reference binding, editing, relighting, outpainting, and scene changes; no repeatability or comparative ranking was inferred.
Banana Pie sells paid access to this model alongside other models in one studio. Our verdicts come from tests run through the same pipeline our users get.