Image GenerationActive

GPT Image 2.5 Flare Custom

Generate new images or edit reference photos with GPT Image 2.5 Flare Custom by OpenAI. Export PNG/JPEG/WebP up to 3840px, including transparent backgrounds.

Text to ImageImage to ImageImage Editing

Model ID

gpt-image-2.5-flare-custom

Provider

openai

Updated

1789043803

wiro playground—openai/gpt-image-2.5-flare-custom

Input Images

Reset to defaults

0 / 16

Optional. Input images to edit, up to 16.

Prompt*

Required. Text description of the desired image or image edit. Maximum 32,000 characters.

Easy Size

Preset

Width*

Required. Output width in pixels (480–3840). Combined limits: long:short ratio at most 3:1, total pixels 655,360–8,294,400. Non-16-aligned values are snapped and resized to your exact request. Falls back to 1024×1024 if no valid size can be found.

Height*

Required. Output height in pixels (480–3840). Combined limits: long:short ratio at most 3:1, total pixels 655,360–8,294,400. Non-16-aligned values are snapped and resized to your exact request. Falls back to 1024×1024 if no valid size can be found.

Sample outputs

Updated 1789043803

Overview

GPT Image 2.5 Flare Custom is an image generation and editing model from OpenAI. It turns a written brief into 1–10 images, or applies targeted edits to an existing image. For edits, you can provide reference images and an optional mask to limit what changes. It’s useful when you need fast, high-quality visuals while keeping key subjects consistent.

What you can build

Product hero shots with consistent materials, logos, and angles
Ad and social variants from one base image (colorways, backgrounds, lighting)
Poster, flyer, and packaging concepts with legible layout and text
UI mockups with readable labels, charts, and tables
Transparent-background cutouts for catalogs, stickers, and thumbnails
Iterative art direction workflows (make one change per round)
Before/after edits for photo retouching and style changes

Inputs

A written description of what you want to generate or how you want to edit the image. It supports long prompts up to 32,000 characters.
Optional reference images to edit or guide the result. Provide up to 16 images total. Use PNG, JPEG, or WebP.
Optional mask image to constrain edits to a region. Provide a single mask image alongside at least 1 reference image.
Output dimensions in pixels, as width and height.
- Both dimensions must be multiples of 16.
- The longest edge can be up to 3840 pixels.
- Aspect ratio must stay between 1:3 and 3:1.
- Total pixel count must be between 655,360 and 8,294,400.
A quality level that trades cost and detail. Options include auto, low, medium, high, extra-high, and max.
A background mode: automatic, forced opaque, or true transparent.
- Transparent output requires a format with alpha support.
Output file format selection: PNG, JPEG, or WebP.
Optional compression level for JPEG or WebP output, from 0 to 100.
Number of images to generate per run, from 1 to 10.
Optional moderation strictness selection. Options typically include auto and low.

Outputs

The model returns one or more generated images. Each result includes an image file in the chosen format (PNG, JPEG, or WebP). You also get basic file metadata for each image, such as content type and pixel dimensions. Workflows that request multiple samples return an ordered list of images.

Recommended settings

Product photos and marketing creatives: use High quality and PNG output for crisp edges and small text.
Posters or layouts with dense typography: use Extra-high or Max quality.
Transparent cutouts: set background to Transparent and export as PNG or WebP.
Fast ideation: use Low or Medium quality, then rerun the best concept at High.
Image editing with a mask: keep the prompt specific about what must not change.

Limitations

Size rules are strict. Invalid dimensions can fail or be adjusted to a valid canvas.
The model is image-only. It doesn’t output video or audio.
Very small text can blur at lower quality levels.
Exact typography, kerning, and perfect spelling aren’t guaranteed.
Edits can still “spill” into nearby regions if the mask edges are ambiguous.
Low-quality inputs increase risk. Blurry photos, heavy compression, screenshots of screenshots, and scanned images can produce artifacts.
Busy or cluttered source images can reduce edit precision.

Safety & compliance

The service applies safety checks to both input images and generated outputs.
Requests may be refused or blocked for disallowed content, including sexual content involving minors, graphic violence, or hateful imagery.
The system is designed to reduce harmful deepfake-style outputs of real people.
Follow OpenAI usage policies and your local laws when generating, editing, or publishing images.
Some accounts may need additional verification to access GPT Image models.

Example prompts

Great starting points for gpt-image-2.5-flare-custom.

Turn this into a dramatic vertical hero shot: the same sneaker floating slightly above a wet black reflective floor, hard rim light from behind, swirling coloured smoke in magenta and teal filling the space around it, keep the shoe's shape, materials and laces exactly as they areImage Generation

Place the same car on a red dirt desert road at sunset, tall rock formations in the distance, long shadow stretching toward the camera, dust hanging in the warm light, deep orange and violet sky above, keep the car's model, colour and angle unchangedImage Generation

Keep the same cat and pose but place it on the surface of the moon wearing a small clear glass space helmet, Earth rising in the black starry sky behind it, cool blue light on its fur from the planet, keep its face, markings and expression identicalImage Generation

Transform the same tower into a futuristic megastructure at dusk: glowing blue accent strips along every floor, holographic signage on the facade, flying vehicles between buildings, dark violet sky with low clouds, keep the building's shape, proportions and camera angle exactly the sameImage Generation

API quick start

Run gpt-image-2.5-flare-custom with a single API call.

POST https://api.wiro.ai/v1/Run/openai/gpt-image-2-5-flare-custom

{
  "prompt": "Turn this into a dramatic vertical hero s…",
  "width": 1024,
  "height": 1024,
  "quality": "low"
}

curl

curl -X POST "https://api.wiro.ai/v1/Run/openai/gpt-image-2-5-flare-custom" \
  -H "Content-Type: application/json" \
  -H "x-api-key: YOUR_WIRO_API_KEY" \
  --data-binary @- <<'JSON'
{
  "prompt": "Turn this into a dramatic vertical hero s…",
  "width": 1024,
  "height": 1024,
  "quality": "low"
}
JSON

View full API docs

Task History

No tasks yet