GPT Image 2
OpenAI's most advanced image model. GPT Image 2 generates and edits images at 1K, 2K, or 4K with three quality levels, native transparent backgrounds, up to 10 input images for editing and reference, and batch output of up to 4 images per run.
50% less than the priciest provider (5 compared). Same model, same output.
Optional images to edit or use as reference (up to 10). Leave empty for pure text-to-image.
Resolution
Output resolution tier.
Aspect Ratio
Aspect ratio of the generated image. 'auto' lets the model choose, and is only available at 1K.
Quality
Rendering quality. Low is a fast draft, high is the premium tier; the price difference between them is large.
Background
Transparent requires a PNG output. Auto lets the model decide.
Output Format
The format of the generated image.
50% less than the priciest provider (5 compared). Same model, same output.
Your result renders here.
Use GPT Image 2 via the API
Generate with GPT Image 2 from your own code. Async by design: submit a generation, get a run id back instantly, poll until it completes. Billed in credits at the best available provider price — from $0.005 per image.
1 — Start a generation
curl -X POST https://app.scenetra.com/api/v1/generate/gpt-image-2 \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"prompt":"a watercolor fox in falling snow, soft light","quality":"medium","aspect_ratio":"1:1"}'Response (202)
{
"runId": "run_abc123",
"status": "PENDING",
"model": "gpt-image-2",
"estimatedCredits": 30,
"statusUrl": "/api/v1/runs/run_abc123"
}2 — Poll until COMPLETED (every 2-5 seconds)
curl https://app.scenetra.com/api/v1/runs/run_abc123 \
-H "Authorization: Bearer YOUR_API_KEY"Response
{
"runId": "run_abc123",
"status": "COMPLETED",
"progress": { "completed": 1, "total": 1 },
"outputs": [
{ "type": "image", "url": "https://images.scenetra.com/..." }
],
"error": null,
"estimatedCredits": 30
}Request fields
| Field | Type | Description | Default |
|---|---|---|---|
| prompt* | string | Describe the image to generate, or the edit to apply to the input images. | — |
| image_urls | string[] (https URLs) | Optional images to edit or use as reference (up to 10). Leave empty for pure text-to-image. | — |
| resolution | string | Output resolution tier.1K · 2K · 4K | 1K |
| aspect_ratio | string | Aspect ratio of the generated image. 'auto' lets the model choose, and is only available at 1K.auto · 21:9 · 16:9 · 3:2 · 4:3 · 5:4 · 1:1 · 4:5 · 3:4 · 2:3 · 9:16 | auto |
| quality | string | Rendering quality. Low is a fast draft, high is the premium tier; the price difference between them is large.low · medium · high | medium |
| background | string | Transparent requires a PNG output. Auto lets the model decide.auto · transparent · opaque | auto |
| num_images | number | How many images to generate (1-4). Billed per image. | 1 |
| output_format | string | The format of the generated image.png · jpeg | png |
| callbackUrl | string (https URL) | Optional. When the run finishes (completed or failed), we POST the final run status to this URL — same shape as the poll response, plus model, params, and chargedCredits. Best-effort; polling remains the source of truth. | |
Unknown fields are rejected with a helpful 400. Media fields take https URLs. There is no provider parameter — every run is routed to the cheapest capable provider automatically.
Good to know
Billing
Runs bill your account credits at provider cost with 0% markup. The submit response includes the credit estimate; failed generations are refunded automatically. Add ?dryRun=1 to get the estimate without running anything.
Errors
Errors are always { "error": { "code", "message" } } with stable codes: unauthorized, model_not_available, invalid_parameter, insufficient_credits, rate_limited. A failed run is not an HTTP error — poll responses report status: "FAILED" with the reason.
What makes GPT Image 2 different
Native transparent backgrounds
GPT Image 2 renders straight to transparency — set Background to transparent (with PNG output) and you get clean cutouts without a separate background-removal step. That's a genuine pipeline shortcut for product shots, stickers, logos, and any asset headed for compositing, and on a Scenetra board the transparent output feeds directly into downstream nodes.
A quality dial with real range
Three rendering levels — low, medium, high — let you match spend to purpose within one node. Low is a fast draft mode from $0.005 per image; medium is the everyday default ($0.03 at 1K); high is the tier for finals, up to $0.08 at 4K. Iterate low, deliver high, all on the same prompt.
Editing and reference with up to 10 images
Attach up to 10 input images and GPT Image 2 becomes an editor: revise a shot, apply a described change, or draw on references for style and subject. OpenAI's instruction-following heritage shows here — complex, multi-clause edit prompts tend to land the way you wrote them.
Up to 4 images per run, 1K to 4K
Generate up to 4 variations in a single run and pick the winner, then re-render at 2K or 4K for delivery. Eleven aspect ratios cover every standard canvas, and the Auto ratio (1K only) lets the model choose framing itself. Everything is billed at provider cost with 0% markup.
Parameters
| Parameter | Type | Description | Default |
|---|---|---|---|
| Prompt* | text | Describe the image to generate, or the edit to apply to the input images. | — |
| Input Images | image upload | Optional images to edit or use as reference (up to 10). Leave empty for pure text-to-image. | — |
| Resolution | select | Output resolution tier.1K · 2K · 4K | 1K |
| Aspect Ratio | select | Aspect ratio of the generated image. Auto lets the model choose, and is only available at 1K.Auto · 21:9 · 16:9 · 3:2 · 4:3 · 5:4 · 1:1 · 4:5 · 3:4 · 2:3 · 9:16 | auto |
| Quality | select | Rendering quality. Low is a fast draft, high is the premium tier — the price difference between them is large.low · medium · high | medium |
| Background | select | Transparent requires a PNG output. Auto lets the model decide.auto · transparent · opaque | auto |
| Number of Images | number | How many images to generate (1-4). | 1 |
| Output Format | select | The format of the generated image.png · jpeg | png |
Pricing
From $0.005 per image
Pay per generation — only pay for what you use.
Low - 1K / 2K
$0.005
per image
Low - 4K
$0.012
per image
Medium - 1K
$0.03
per image
Medium - 2K
$0.042
per image
Medium - 4K
$0.101
per image
High - 1K
$0.03
per image
High - 2K
$0.05
per image
High - 4K
$0.08
per image
Use cases
Marketing & ads
- Campaign visuals up to 4K
- Transparent logo and badge assets
- Quality-tiered draft-to-final workflow
- 4-variant batches for A/B tests
E-commerce
- Product cutouts on transparency
- Listing edits from reference shots
- Composite scenes from up to 10 inputs
- Consistent catalog series
Product & UI design
- Icons and stickers with alpha channels
- Mockup imagery at precise ratios
- Asset variants per platform
- High-quality finals at 4K
Social content
- Vertical 9:16 and 4:5 posts
- Thumbnails in batches of 4
- Overlay graphics on transparency
- Fast drafts at low quality
Frequently asked questions
What is GPT Image 2?+
GPT Image 2 is OpenAI's most advanced image generation model. It creates images from text and edits from up to 10 input images, at 1K, 2K, or 4K resolution with three quality levels, native transparent backgrounds, and up to 4 images per run. On Scenetra it runs as a node in the visual workflow editor.
How much does GPT Image 2 cost?+
From $0.005 per image on Scenetra. Low quality runs $0.005 at 1K or 2K and $0.012 at 4K; medium is $0.03 at 1K, $0.042 at 2K, and $0.101 at 4K; high is $0.03 at 1K, $0.05 at 2K, and $0.08 at 4K. The node shows the exact price for your chosen settings before you run, and billing is per generation at provider cost with 0% markup.
Can GPT Image 2 generate transparent backgrounds?+
Yes — transparency is a native output option. Set Background to transparent and Output Format to PNG, and the model renders directly to an alpha channel with no separate cutout step.
Can GPT Image 2 edit existing images?+
Yes. Attach up to 10 input images and describe the edit or use them as style and subject references. On a Scenetra board, inputs can also be the outputs of other nodes, so GPT Image 2 can serve as an editing stage in a longer pipeline.
What do the quality levels change?+
Quality controls how much rendering effort the model spends. Low is a fast draft tier from $0.005 per image, medium is the balanced default, and high is the premium tier for final assets. A common workflow is iterating at low and re-running the final prompt at medium or high.
What resolutions and aspect ratios are supported?+
Three resolutions — 1K, 2K, and 4K — and eleven aspect ratios from 21:9 ultrawide to 9:16 vertical. The Auto ratio, where the model picks the framing itself, is available at 1K only.
Do I need an OpenAI API key to use GPT Image 2 on Scenetra?+
No. Scenetra handles the infrastructure — you sign up, drop a GPT Image 2 node on a board, and generate. It's pay-per-generation from your credit balance with no subscription required.
One-click tools built on GPT Image 2
Ready-made workflows — fill in a form, no node editor needed.
AI Fashion Photoshoot Generator
Upload your clothing photos and get a full AI fashion photoshoot — a generated model wears your garments across styled, cinematic shots.
Open tool →AI Character Video Generator
Describe a character in text and get four videos of them — street interview, living-room dance, kitchen talk, and cooking — with generated audio and a consistent identity.
Open tool →AI Product Ad Generator
Turn one product photo into four scroll-stopping animated ad clips — background removed, product staged in bold scenes, then brought to life with video.
Open tool →Consistent Character Generator
Turn one face photo into a 12-angle, identity-locked character reference grid — profiles, expressions, and close-ups of the same person.
Open tool →AI Character Creator
Create a complete AI character from a text description — casting portrait, upscaled master face, and identity-locked reference sheets for angles, emotions, hair, wardrobe, and body.
Open tool →AI UGC Video Generator
Turn one face photo into four ready-to-post UGC-style videos — street interview, living-room dance, kitchen talk, and garden chat — starring a consistent AI creator.
Open tool →AI Headshot Generator
Turn one selfie into 16 professional headshots — LinkedIn, studio portrait, editorial, and creative looks in a single run.
Open tool →Related models
GPT Image 2.5 Sunburst
OpenAI's precision image model. GPT Image 2.5 Sunburst generates and edits images at 1K, 2K, or 4K with five quality levels up to Max, native transparent backgrounds, up to 16 reference images for exact, instruction-following edits, and up to 4 images per run.
View model →Nano Banana 2
Generate high-quality images at up to 4K resolution using Google's Gemini 3.1 Flash model. Nano Banana 2 is optimized for speed, supports text-to-image and image editing with reference images, and offers flexible aspect ratios.
View model →Seedream 5 Pro
ByteDance's most advanced image generation and editing model. Seedream 5 Pro delivers high-fidelity output at 1K (about 2MP) or 2K (about 4MP), supports up to 10 reference images for editing and composition, and covers nine aspect ratios from ultrawide 21:9 to vertical 9:16 — including an Auto mode that infers framing from your prompt.
View model →Flux 2 Pro
Black Forest Labs' professional-grade image model delivering high-quality outputs with precise control. Flux 2 Pro supports text-to-image generation and image editing with flexible sizing, safety controls, and simple flat per-image pricing.
View model →Start creating with GPT Image 2
Use GPT Image 2 alongside 50+ other AI models in Scenetra's visual workflow editor. No setup required.
Get Started