OpenAI
Operational

GPT Image 2 API

GPT Image 2 API: OpenAI's image model for text-heavy visuals at $0.02 per image.

Price per image62%off
1K
$0.02$0.053
2K
$0.02$0.212
4K
$0.02$0.419
Get your API keyTry it in the playground
Model IDgpt-image-2
generate.py
from openai import OpenAI
client = OpenAI(
    base_url="https://api.zurelay.com/v1",
    api_key="YOUR_ZURELAY_KEY",
)
image = client.images.generate(
    model="gpt-image-2",
    prompt="A red fox sitting in fresh snow at dawn, photo",
    size="1536x1024",
    quality="high",
)
print(image.data[0].url)

Base URL https://api.zurelay.com/v1. Works with any OpenAI SDK.

Sizes
Up to 3840 px
Price
Same at any size
Input
Text, Image
Output
Image
Released
Apr 21, 2026
Uptime, 7 days
New

Pricing

GPT Image 2 API pricing. A fraction of the list price.

Pay as you go from prepaid credit, with no subscription and no minimum. Requests that fail are never billed.

RatezurelayOpenAI listYou save
1K image
1024 × 1024, any quality
$0.02$0.05362%off
2K image
2048 × 2048, any quality
$0.02≈$0.21291%off
4K image
3840 × 2160, any quality
$0.02≈$0.41995%off

One price at every size, always high quality. OpenAI bills images by output tokens, which grow with the image's area, so its 2K and 4K prices (≈) are worked out from its 1024 × 1024 price.

Savings calculator

Move the sliders to your monthly usage.

Images per month1,000
Resolution
zurelay
$20.00
OpenAI list price
$53.00
You save every month$33.00

$396.00 a year

Overview

What is GPT Image 2?

GPT Image 2 is OpenAI's image generation and editing model for production visuals such as infographics, posters and text-heavy layouts, released on April 21, 2026. The GPT Image 2 API on zurelay turns a text prompt into images through one OpenAI-compatible endpoint. Every image is made at high quality, at one flat price for any size up to 3840 pixels.

GPT Image 2 is the image model OpenAI released on April 21, 2026, for its API and Codex. OpenAI describes it as a model for fast, high-quality image generation and editing, built for production work where images must be accurate, readable, on-brand and formatted for the surface they appear on. It brought stronger structured generation for diagrams, infographics, charts, posters and comics, plus improved multilingual text rendering. The GPT Image 2 API on zurelay uses the model ID gpt-image-2 and also accepts openai/gpt-image-2.

GPT Image 2 takes any size written as WIDTHxHEIGHT that meets OpenAI's rules: both edges are multiples of 16, no edge passes 3840 pixels, the long edge is at most three times the short edge, and the total falls between 655,360 and 8,294,400 pixels. Popular sizes include 1024x1024, 1536x1024, 1024x1536, 2048x2048, 2048x1152 and 3840x2160, and square images are usually the fastest to generate. OpenAI treats outputs above 2560x1440 as experimental, so 4K results can vary more. Quality can be low, medium, high or auto, output comes as PNG, JPEG or WebP, and transparent backgrounds are in preview with PNG or WebP.

OpenAI followed GPT Image 2 with GPT Image 2.5 Flare and Sunburst on September 8, 2026. It says Flare cuts latency by 50% compared with GPT Image 2 at comparable image quality, and that Sunburst's image quality is higher than GPT Image 2's. OpenAI now points new integrations to the 2.5 models and keeps GPT Image 2 documented for existing workflows. OpenAI's GPT Image 2 can also edit images from image inputs, but zurelay offers it for text-to-image generation only.

OpenAI bills GPT Image 2 by tokens for text input, image input and image output, so its cost per image depends on the quality and size you pick. At 1024x1024, its high quality costs about four times its medium quality. zurelay makes every image at high quality and charges one flat price at any size up to 3840 pixels, so a 4K image costs the same as a square. Send requests to POST https://api.zurelay.com/v1/images/generations in the OpenAI images format and get a URL back by default, or base64 with response_format set to b64_json.

Strengths

Where GPT Image 2 shines. And what teams build with it.

01

Made for text-heavy visuals

OpenAI built GPT Image 2 for diagrams, infographics, charts, posters and comics. It also improved text rendering across languages.

02

Sizes up to 4K

Use any size with edges in multiples of 16 and a ratio up to 3:1, up to 3840 pixels on the long edge. OpenAI marks outputs above 2560x1440 as experimental.

03

One price at any setting

zurelay bills a flat price per image and makes every image at high quality. Larger sizes don't raise your cost.

04

Follows detailed prompts

OpenAI reports more reliable instruction following, detail preservation and composition in GPT Image 2. It says that means more usable images with less cleanup.

Use cases

  • Infographics and diagrams

    Explainers, labeled diagrams, charts and timelines with text inside the image. Check labels and spelling before you publish.

  • Posters and ads in any language

    Campaign graphics, posters and social posts with in-image text in the language your audience reads.

  • Assets for every surface

    Wide banners, tall story frames and 4K wallpapers from one model, using custom sizes up to a 3:1 ratio.

  • High-volume generation

    Product visuals, thumbnails and variation sets where a flat price per image keeps the bill predictable.

Get started

Call GPT Image 2 in three steps.

No waitlist and no new SDK. If your code already talks to OpenAI, it already talks to zurelay.

  1. 1

    Create a key

    Sign up, add credit and create an API key. Set a monthly budget or a rate limit per key if you like.

    Get your API key
  2. 2

    Point your SDK at zurelay

    Change the base URL. Everything else in your code stays the same.

    https://api.zurelay.com/v1
  3. 3

    Use gpt-image-2

    Send a prompt to /v1/images/generations with a size and aspect ratio, and get back a link to your image.

    gpt-image-2

Works with the tools you already use

Compare

Similar models. All on the same key.

62%off

GPT Image 2.5 Flare

$0.02 per image

Pick GPT Image 2.5 Flare when speed matters, since OpenAI says it cuts latency by 50% compared with GPT Image 2 at comparable quality.

GPT Image 2.5 Flare API
62%off

GPT Image 2.5 Sunburst

$0.02 per image

Pick GPT Image 2.5 Sunburst when a prompt needs more detail than GPT Image 2 delivers, since OpenAI rates its image quality higher.

GPT Image 2.5 Sunburst API
74%off

Nano Banana Pro

from $0.035 per image

Pick Nano Banana Pro to compare Google's image model, with fixed 1K, 2K and 4K outputs and a focus on legible text.

Nano Banana Pro API

FAQ

GPT Image 2 API questions.

Anything else, write to sales@zurelay.com and an engineer will answer.

How much does the GPT Image 2 API cost per image?

zurelay charges $0.02 per image for gpt-image-2 at every size up to 3840 pixels, and every image is made at high quality. OpenAI bills by tokens, and its cost for a 1024x1024 image at medium quality is about $0.053, so you save 62%. OpenAI's high quality costs about four times its medium price at that size, so the gap is wider for high-quality work.

Is there a cheaper GPT Image 2 API than OpenAI's?

Yes. zurelay serves gpt-image-2 at a flat $0.02 per image, 62% below OpenAI's approximate cost for a 1024x1024 medium-quality image. zurelay's price stays the same at every size, and every image comes back at high quality rather than medium.

How do I call the GPT Image 2 API with the OpenAI SDK?

Create an OpenAI client with the base URL set to https://api.zurelay.com/v1 and your zurelay API key. Call images.generate with model set to gpt-image-2, a prompt and a size in pixels such as 1536x1024. You can also pass output_format and n (1 to 4). The response holds an image URL by default; set response_format to b64_json to get base64 instead.

What sizes does GPT Image 2 support, and is there an aspect_ratio setting?

Pass size as WIDTHxHEIGHT in pixels, such as 1024x1024, 1536x1024, 2048x1152 or 3840x2160; there is no aspect_ratio parameter for this model. Both edges must be multiples of 16, no edge may pass 3840 pixels, the long edge can be at most three times the short one, and the total must fall between 655,360 and 8,294,400 pixels. OpenAI treats outputs above 2560x1440 as experimental. On zurelay every image is made at high quality, whatever quality you pass.

GPT Image 2 vs GPT Image 2.5: which should I use?

OpenAI says GPT Image 2.5 Flare cuts latency by 50% compared with GPT Image 2 at comparable quality, and rates Sunburst's image quality above GPT Image 2. If a GPT Image 2 workflow already meets your quality bar, OpenAI suggests testing Flare for speed, and Sunburst when GPT Image 2 falls short. On zurelay all three cost the same flat $0.02 per image, so you can compare them on your own prompts.

How good is GPT Image 2 at rendering text?

OpenAI built GPT Image 2 for text-heavy visuals such as diagrams, infographics and posters, with improved multilingual text rendering. For exact wording, put the text in quotes or capitals, give its font, size, color and placement, and spell brand names letter by letter. OpenAI notes the model can still struggle with precise text placement, so check spelling before you publish.

How long does zurelay keep generated images?

Images are stored for 30 days behind the URL in the response. Download any image you want to keep before then. If you request b64_json, the image data comes back in the response itself.

Is this the same GPT Image 2 model as OpenAI's?

Yes. zurelay runs the same gpt-image-2 model and accepts requests in the OpenAI images format. Image generation varies from run to run, so the same prompt will not return byte-identical images. zurelay is an independent service and is not affiliated with OpenAI.

Can I edit images or use reference images with GPT Image 2 on zurelay?

No. OpenAI's GPT Image 2 accepts image inputs for edits and inpainting, but zurelay offers text-to-image generation only, through /v1/images/generations. Image edits, masks and reference-image inputs are not available.

More models on zurelay

All models

Stop paying list price.
Start saving today.

Create a key in seconds, point the SDK you already use at zurelay, and every request costs up to 85% less from the first token.

Get your API keyTalk to an engineer