GPT Image 2 generated hero artwork

GPT Image 2

One model. Every image workflow.

Create new visuals, transform existing assets and keep important reference details intact. GPT Image 2 brings fast, high-quality generation and editing to DreamAPI — with flexible sizing from social crops to 4K.

View API Docs ↗Try in Playground →
The workflow

From blank prompt
to precise revision.

Generate campaign-ready imagery from text, then carry the visual forward through reference-driven edits without rebuilding the scene.

01

Generate from text

Create campaign-ready imagery from a natural-language prompt.

02

Edit specific details

Transform assets while preserving subject, lighting and composition.

03

Iterate without rebuilds

Carry the visual forward through reference-driven edits.

Capabilities

One model.
Every image workflow.

GPT Image 2 covers generation, editing and flexible sizing — fast, high quality, and tuned for precise creative control.

Editing specific details

GPT Image 2 automatically processes input images at high fidelity. Preserve the subject, lighting and composition while changing only the detail the prompt asks for.

Building the described image

Use natural-language prompts to specify composition, material, lighting, layout and exact creative constraints. GPT Image 2 is designed for high-quality results without turning every brief into a prompt-engineering project.

Sizing for the placement

Go beyond fixed presets. GPT Image 2 accepts thousands of valid resolutions, with a maximum edge of 3840 pixels and aspect ratios up to 3:1 — useful for product pages, ads, social formats and high-resolution campaign work.

Reference images

Bring more than a prompt.

Guide generation with reference images and precise edits, then keep the details that matter in every output.

Original image before a specific-detail edit
Before
Image after a specific-detail edit
After
Precise edits

Editing specific details with ease.

Preserve the subject, lighting and composition while changing only the detail the prompt asks for.

High fidelityPreserve subjectDetail only
GPT Image 2 gallery example 1
GPT Image 2 gallery example 2
GPT Image 2 gallery example 3
GPT Image 2 gallery example 4
GPT Image 2 gallery example 5
GPT Image 2 gallery example 6
Reference generation

Build the image you described.

Specify composition, material, lighting, layout and exact creative constraints in plain language.

CompositionMaterialLighting
Flexible sizing

Size the output
for the placement.

Go beyond fixed presets — from social crops to high-resolution campaign work, GPT Image 2 accepts thousands of valid resolutions with a maximum edge of 3840 pixels and aspect ratios up to 3:1.

Output constraintsSquare 1:1 · Landscape 16:9 · Portrait 3:4 — max edge 3840 px, aspect ratios up to 3:1.
Square 1 to 1 GPT Image 2 output
Landscape 16 to 9 GPT Image 2 output
Portrait 3 to 4 GPT Image 2 output
Developer API

API now
available.

Configure GPT Image 2 inside DreamAPI for generation and editing workflows. The live API workspace provides the current endpoint, authentication and response schema.

Text → ImageReference → ImageImage → EditMulti-turn workflow
01

Choose a mode

Generate from text, edit an existing image, or run a multi-turn workflow.

02

Set quality & size

Pick low, medium or high quality and any supported resolution up to a 3840px edge.

03

Generate or edit

Return PNG, JPEG or WebP output through the DreamAPI image endpoint.

Model IDgpt-image-2
Qualitylow / medium / high
Popular output1K / 2K / 4K
Format controlsPNG / JPEG / WebP
View API documentation →
generate
const image = await dreamapi.images.generate({
  model: "gpt-image-2",
  prompt: "A premium studio product campaign...",
  size: "2048x2048",
  quality: "high",
  output_format: "png",
  n: 1
});
curl https://api.dreamapi.com/v1/images/generations \
  -H "Authorization: Bearer $DREAMAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-image-2",
    "prompt": "A premium studio product campaign...",
    "size": "2048x2048",
    "quality": "high",
    "output_format": "png",
    "n": 1
  }'
Choose the right image engine

GPT Image 2 vs the field.

A workflow-focused snapshot of GPT Image 2, Nano Banana Pro and Seedream 5 Pro.

Model comparison

DimensionGPT Image 2Nano Banana ProSeedream 5 Pro
Best fitPrecise text, constraints and iterative editingPhotoreal, balanced, fast creative workControllable, iterative, cost-efficient production
ResolutionFlexible sizes up to a 3840 px edgeNative 4K, up to 4096 × 4096Native 2K; 4K typically enhanced
Text accuracyApprox. 98.5–99%Approx. 94–96%Approx. 89–90%; strong multilingual layouts
EditingMulti-turn editing with contextual continuityRegion editing with broader granularityPoint, box, lasso and sketch editing
Image referencesAlways processed at high fidelityUp to 14 imagesApprox. 5–14 depending on provider
Layer exportFlattened outputFlattened output10+ transparent PNG layers reported
PhotorealismStrong realism with reduced artificial textureLeading skin, fabric and material realismStrong composition and complex spatial logic
Speed profileFast low-quality drafts; slower at high qualityFast, with longer 4K outputModerate
Workflow choiceText-heavy ads, UI, brand layouts, strict briefsFashion, product photography, realistic scenesLayered design, local edits, budget-sensitive scale

Directional comparison based on the previously prepared DreamAPI model research. Capabilities, provider limits and benchmark results may change.

Pricing plans

Top up credits.
Pay as you generate.

One credit pool for every model on DreamAPI — video, image, avatar and editing. Buy a pack, then spend credits at the per-model rates below.

Credit packs are purchased on the DreamAPI pricing page.

DreamAPI billing

GPT Image 2 pricing.

Credits are calculated from total pixel count, quality and image quantity. The values below reproduce the billing table supplied for DreamAPI.

GPT Image 2 — credits per image

ModelResolutionLowMediumHigh
gpt-image-21024 × 102421872
gpt-image-22048 × 2048436144
gpt-image-22880 × 2880654216
Get API key →

Credits per generated image. Multiply the result by image count. credits = sizeBase × qualityMultiplier × n

GPT Image 2 API FAQ

Before your
first image.

What is GPT Image 2?+

GPT Image 2 is OpenAI's state-of-the-art model for fast, high-quality image generation and editing. It accepts text prompts and image inputs, and produces image outputs.

Can GPT Image 2 edit existing images?+

Yes. It supports full-image edits, prompt-guided masked edits and iterative image workflows. Input images are processed at high fidelity automatically.

What image sizes can I request?+

DreamAPI pricing examples cover 1024², 2048² and 2880² outputs. The underlying model supports flexible dimensions up to a 3840-pixel maximum edge, subject to its pixel and aspect-ratio constraints.

Which quality levels are available?+

Use low for fast drafts, medium for balanced production work and high for final assets. DreamAPI credits scale with the selected quality and pixel tier.

Which formats are supported?+

GPT Image 2 can return PNG, JPEG or WebP. JPEG and WebP also support configurable compression.

Does GPT Image 2 support transparent backgrounds?+

No. The current GPT Image 2 API does not support transparent backgrounds.

Now live on DreamAPI

Make the image.
Keep the intent.

Generate, edit and iterate with GPT Image 2 through one production-ready API and one DreamAPI credit balance.

Try GPT Image 2 →Read the docs