GPT-Image-1 — OpenAI's Image Generation Model 📝❓🖼️🔢 → 🖼️

⭐ Official ▶️ 1.8M runs 📅 Apr 2025 ⚙️ Cog 0.16.7 📄 Paper ⚖️ License
image-editing image-to-image text-to-image image-generation openai

Performance

45.9sTypical run time
1.8MTotal runs

Compute billed by the provider (bring your own API key)

Data sampled 2026-07-12

GPT-Image-1 is OpenAI's image generation model, available on Replicate for quick testing and API access. Note that it requires your own verified OpenAI API key and bills usage to your OpenAI account.

About

GPT-Image-1 is OpenAI's dedicated image generation model, built to produce high-quality visuals from text prompts. It handles photorealistic scenes, illustrations, diagrams, and — unusually for an image model — text rendered inside the image, which makes it a strong fit for posters, ad creative, UI mockups, and infographics.

A key thing to know before you start: on Replicate this model runs as a passthrough to OpenAI. You must supply your own verified OpenAI API key (openai_api_key is a required input), and usage is billed to your OpenAI account rather than by Replicate GPU time.

How it compares to other image models on Replicate

Model Type Key strength Billing model
openai/gpt-image-1 Text-to-image + image editing Accurate in-image text, strong prompt adherence, transparent backgrounds Your own OpenAI key (billed by OpenAI)
black-forest-labs/flux-1.1-pro Text-to-image Fast, photorealistic scenes Per run on Replicate
stability-ai/sdxl Text-to-image Open weights, large LoRA / fine-tune ecosystem Per run on Replicate
google/imagen-4 Text-to-image Photorealistic detail and typography Per run on Replicate

Key inputs that matter

  • prompt (required) — supports long, detailed, multi-part descriptions.
  • openai_api_key (required) — bring your own verified OpenAI key.
  • input_images — pass existing images to edit or compose from, not just generate from scratch.
  • background — supports transparent backgrounds, useful for logos, stickers, and overlays.
  • aspect_ratio / quality / number_of_images — control shape, fidelity, and batch size.
  • moderation — OpenAI's content filtering is applied and configurable here.

When to use this model

  • Images that need legible text — posters, ads, slides, infographics.
  • Editing or compositing existing images via input_images.
  • Transparent-background assets for design work.
  • Prompts with complex spatial relationships or multiple objects.

When to reach for something else

  • You don't want to manage a verified OpenAI account and key — try flux-1.1-pro or SDXL.
  • You need custom styles via LoRA / fine-tuning — SDXL's ecosystem is far larger.
  • You need the lowest latency — generation here takes tens of seconds per image.

Limitations and gotchas

  • Requires a verified OpenAI account; new or unverified keys are rejected.
  • Usage is billed by OpenAI, so Replicate's per-run pricing does not apply.
  • Content moderation is enforced and cannot be fully disabled.
  • Not open weights — no LoRA, no self-hosted fine-tuning.

Example Output

Prompt:

"Add the floral pattern to the vase"

Output

Example output

Performance Metrics

45.87s Prediction Time
45.89s Total Time
1 Images
All Input Parameters 
{
  "prompt": "Add the floral pattern to the vase",
  "quality": "auto",
  "background": "auto",
  "moderation": "auto",
  "aspect_ratio": "1:1",
  "input_images": [
    "https://replicate.delivery/pbxt/MusWuQJm1RJPu1Cj0ajRmoMnHyYNPk6ljT1QCU4DbHMsqDTF/53541851-62f3-44a7-b075-ef053ae2f324.jpg",
    "https://replicate.delivery/pbxt/MusWuPkfcvyZQuuXeMIbQXEMe9K2G8rDCNraQffAt0OzMRaT/colored-flower-pattern-free-vector.jpg"
  ],
  "output_format": "webp",
  "openai_api_key": "[REDACTED]",
  "number_of_images": 1,
  "output_compression": 90
}
Input Parameters
prompt (required) Type: string
A text description of the desired image
quality Default: auto
The quality of the generated image
user_id Type: string
An optional unique identifier representing your end-user. This helps OpenAI monitor and detect abuse.
background Default: auto
Set whether the background is transparent or opaque or choose automatically
moderation Default: auto
Content moderation level
aspect_ratio Default: 1:1
The aspect ratio of the generated image
input_images Type: array
A list of images to use as input for the generation
output_format Default: webp
Output format
input_fidelity Default: low
Control how much effort the model will exert to match the style and features, especially facial features, of input images
openai_api_key (required) Type: string
Your OpenAI API key
number_of_images Type: integerDefault: 1Range: 1 - 10
Number of images to generate (1-10)
output_compression Type: integerDefault: 90Range: 0 - 100
Compression level (0-100%)
Output Schema

Output

Type: array • Items Type: string • Items Format: uri

Example Execution Logs 
Generating image...
Image size: 378.3KB
Image size: 303.0KB
Generated image in 45.6sec
Generated 1 images
Input text token count: 711
Input image token count: 646
Output token count: 4160
Version Details
Version ID
5ac56c15446a60fa63b3823de926ada90f5971c2cf9b1dd07659126cfda434e6
Version Created
September 22, 2025
Run on Replicate →