GPT Image 2
OpenAI GPT Image 2 follows long, detailed instructions and composes scenes with several subjects; priced per quality level and output size.
Try it
You can fill this in now — signing in happens when you run it, and your inputs come with you.
At a glance
- Developer
- OpenAI
- Can do
- Text to image, Image to image
- Credits per run
- 1 credits
- Typical time
- ~15s
- Added
- 2026-08-11
Parameters
What the model takes. The form above is generated from exactly this list, and the API accepts the same field names.
| Parameter | Type | Default | Range / options |
|---|---|---|---|
image_urlsrequiredReference images | Image upload | — | ≤ 5 |
promptPrompt | Long text | — | — |
aspect_ratioAspect ratio | Choice | 1:1 | 1:1 · 3:2 · 2:3 · 4:3 · 3:4 · 16:9 · 9:16 · 21:9 |
resolutionResolution | Choice | 2K | 1K · 2K · 4K |
qualityQuality | Choice | low | Low · Medium · High |
num_imagesNumber of images | Choice | 1 | 1 · 2 · 3 · 4 |
Use it from the API
The same run, as a request. Send your API key as a bearer token; poll the returned request_uid until it is done.
curl -X POST https://api.wisaven.com/v1/ai/async/tasks \
-H "Authorization: Bearer $WISAVEN_API_KEY" \
-H "X-App-Id: wisaven" -H "Content-Type: application/json" \
-d '{
"model": "gpt-image-2",
"task": "text-to-image",
"input": {
"image_urls": "https://…",
"prompt": "your prompt here",
"aspect_ratio": "1:1",
"resolution": "2K"
}
}'Questions
What can GPT Image 2 do?
Text to image, Image to image. Fill in the parameters on this page to try it.
How many credits does GPT Image 2 cost?
About 1 credits per run, charged at the platform's live price when you run it. One credit balance covers every workspace.
How long does it take?
Around 15 seconds; heavier settings or a queue take longer, and the page shows progress.
Do I need an account, and who owns the output?
Browsing and filling in the form need no account; running it does (new accounts start with free credits). What you generate is yours, personally or commercially, subject to the terms.
Other models like this
Where this model is used: Image·All models