OpenAIOpenAI Image generation and reference-led editing

GPT Image 2

Create reviewable product edits, campaign layouts, and placement-specific images.

OpenAI documents GPT Image 2 for image generation and editing with flexible sizes. Use the standard TokenHot route when you want to evaluate the exact model ID already configured in your image workflow.

  • Text to image
  • Optional Playground reference
  • Per-image pricing

The published example uses a text prompt with the Images generation endpoint. Attaching a reference in the Playground sends an Images edit request with that one file. Verify the exact route and fields before integrating edits.

Try this model

Choose an aspect ratio and resolution to generate an image.

gpt-image-2

Input

Reference image (optional)

JPG / PNG · Up to 10 MB. Uploaded when you generate.

Aspect ratio
Resolution

1024 × 1024 px

Uses your account balance at the model’s current rates.

Generated result

Your creation appears here

Enter a prompt, choose your settings, and generate.

USD /image

Pricing

GroupPrice
default$0.028
default
$0.028

Public list prices are shown in USD. Final charges may vary by account group and usage tier.

Overview

What is GPT Image 2?

The standard TokenHot route has its own model ID and per-image price row. OpenAI documents GPT Image 2 for generation and reference-based editing.

Use the Playground to choose a ratio and size for each candidate. Leave the reference blank to start from a prompt, or attach a JPG or PNG to guide an edit. Review the image at the intended placement before accepting it.

Generation

Describe one deliverable at a time

Give the subject, framing, and placement for each candidate.

Reference editing

Name the details that should stay

Attach one approved source image in the Playground and describe the change separately from the details to preserve.

Output sizing

Check the target placement

The Playground offers five aspect ratios and 1K, 2K, or 4K size selections. Check the resulting pixel dimensions in your workflow.

From brief to deliverable

Put GPT Image 2 to work

Start with a concrete task and decide what a useful result looks like.

Product imaging teams

Change a product surface without changing the product

Use a reference photo to request a different surface under the same bottle. Compare the candidate against the source before accepting it.

Tools you need
The Playground reference input and side-by-side review.
Keep in mind
Check the cap and silhouette at full size. A requested preservation constraint can still drift.
View a task brief

Keep this bottle's shape, cap, viewing angle, and placement. Replace the white surface with matte slate and preserve a realistic contact shadow. Leave the bottle unbranded.

Campaign design teams

Proof a campaign poster before export

Draft one poster around a short approved headline. Check contrast, copy, and margins before rebuilding the final typography if needed.

Tools you need
The Playground and a layout editor.
Keep in mind
Treat generated lettering as draft. Proof each character against approved copy.
View a task brief

Create a poster draft using the supplied short headline exactly. Make the focal subject clear, use one orange accent, and keep the headline within the supplied print-safe area.

Ecommerce design teams

Recompose a listing image for a wider hero

Use the same desk lamp reference in two requests. Ask for a centered square listing and an offset wide version with room for headline copy.

Tools you need
The Playground ratio selector and real page mockups.
Keep in mind
Review both candidates independently. Switching ratios can alter framing and subject detail.
View a task brief

Use the supplied lamp reference. Keep the lamp shape, material and finish. For this request, follow the selected placement: center the lamp for a square listing or place it to one side of a wide hero with room for headline copy.

Capabilities in context

Key features

FeatureWhat you getWhy it matters
Model IDgpt-image-2Use the exact ID for this TokenHot route.
Generation examplePOST /v1/images/generationsThe public example sends a text prompt and a 1024 × 1024 size.
Playground referenceOne optional JPG or PNG, up to 10 MBThe client submits an edit request when you attach the file.
Playground size choices1K, 2K, and 4KPixel dimensions depend on the chosen ratio.
Billing displayUSD per imageUse the current TokenHot price table for the selected route.

OpenAI documents the underlying model. TokenHot's page and Playground describe this route. The scene illustrations were made with ImageGen to explain the workflows. OpenAI GPT Image 2 model · TokenHot model and API example · TokenHot GPT Image 2 generation guide

Audience & fit

Where GPT Image 2 fits

Product imaging teams

A candidate bottle image with a revised surface.

Campaign design teams

A poster direction with copy and spacing proof notes.

Ecommerce design teams

Two placement candidates based on the same product.

Cost context

Plan the cost of a useful result

Read the live route price

Use the selected model ID and current TokenHot price table for estimates.

Count rejected drafts

Include retries and separately generated variants in the cost of the set.

Compare cost per accepted image

Track how many candidates pass review and how much cleanup is needed before choosing a route.

Use the current pricing table for available rates and account groups. Model IDs and prices come from TokenHot’s catalog; provider pricing and subscriptions are separate.

API

Code examples

curl --location --request POST 'https://api.tokenhot.ai/v1/images/generations' \
--header 'Authorization: Bearer <token>' \
--header 'Content-Type: application/json' \
--data-raw '{
    "model": "gpt-image-2",
    "prompt": "A cute baby sea otter",
    "size": "1024x1024"
}'

Blog

Related Articles

FAQ

Frequently asked questions

What is GPT Image 2 best suited for?

GPT Image 2 is suited to the input, output, and capability types listed on this page. Test production workloads before choosing it for a critical system.

How is GPT Image 2 priced?

Pricing is shown in the pricing section above. Actual cost depends on usage volume, account group, and the request payload.

How can I call GPT Image 2?

Use the model ID shown above with one of the supported API protocols. Code examples are provided when usage data is available.

Which model ID should I use for GPT Image 2?

Use gpt-image-2 in the TokenHot request.

Which endpoint does the public code example use?

It sends a text prompt to POST /v1/images/generations with a size of 1024x1024.

Can I attach a reference in the Playground?

The Playground accepts one optional JPG or PNG up to 10 MB. With the file attached, the client sends an image-edit request. Confirm the fields required by your API integration.

Is Sale a separate OpenAI model?

The current TokenHot catalog describes Sale as a discounted route to GPT Image 2 with its own TokenHot model ID. Check both live price tables and validate the operations you use.

How do I estimate a batch cost?

Use the current per-image price for your route and count all candidates, including alternatives and retries.

Get started

Build with GPT Image 2

Use in Console