Models for Images | OpenAI | Page 1

latent dagger Apr 18, 2026, 2:44 AM

#

Which models for image generation are currently supported via ChatGPT and API?

narrow sparrowBOT Apr 18, 2026, 2:44 AM

#

Hi @latent dagger. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:

#

From the provided sources, here’s what’s explicitly documented.

In the API

You can generate images through:

1. GPT Image models (recommended)

Supported in both the Image API and as the image tool in the Responses API:

gpt-image-1.5 – latest, state‑of‑the‑art GPT Image model [GPT Image 1.5; Image overview]
gpt-image-1
gpt-image-1-mini [Model comparison]

These three are supported by the Image API’s generations/edits endpoints and by the Responses API’s built‑in image_generation tool. [Image overview]

When using the Responses API, you do not set model to a GPT Image model. Instead, you use a text model (e.g. gpt-4.1, gpt-5, etc.) plus the hosted image tool, which internally calls one of the GPT Image models. [Tools image-generation models]

Mainline models that can call the image tool include:

gpt-4o, gpt-4o-mini
gpt-4.1, gpt-4.1-mini, gpt-4.1-nano
o3
gpt-5
gpt-5.4, gpt-5.4-mini, gpt-5.4-nano
gpt-5-nano
gpt-5.2 [Tools image-generation models]

2. DALL·E models (deprecated but still available)

Via the Image API:

dall-e-2
dall-e-3 [Image overview; DALL·E 3 model]

#

Both DALL·E 2 and 3 are marked as previous/older generation and are scheduled to be discontinued on 2026‑05‑12; GPT Image is recommended instead. [Model comparison]

In ChatGPT

The sources state:

GPT‑4o image generation is “the default image generator in ChatGPT.” [ChatGPT images]
Users can still access DALL·E via a dedicated “DALL·E GPT” inside ChatGPT. [ChatGPT images]

Separately, the chatgpt-image-latest model is documented as “Image model used in ChatGPT,” exposed via the same endpoints as GPT Image 1.5 (including image generations and edits). [chatgpt-image-latest]

The documents do not spell out exactly how GPT‑4o and chatgpt-image-latest relate internally, only that:

GPT‑4o is the default image generator in the ChatGPT product. [ChatGPT images]
chatgpt-image-latest is “the image snapshot currently used in ChatGPT” and is available via the API. [chatgpt-image-latest]

So, based strictly on these sources:

#

ChatGPT: GPT‑4o (default image generator) and a DALL·E GPT; internally uses the chatgpt-image-latest image model.
API: GPT Image models (gpt-image-1.5, gpt-image-1, gpt-image-1-mini), plus dall-e-2 and dall-e-3 via the Image API, and a set of text models that can call the image tool in the Responses API.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@latent dagger Kapa.ai is still learning and improving, please let me know how I did by reacting below.

latent dagger Apr 18, 2026, 2:45 AM

#

@narrow sparrow is GPT 4o GPT Image 1?

narrow sparrowBOT Apr 18, 2026, 2:45 AM

#

Let me check the knowledge base for you again. One moment! :books:

#

No. GPT‑4o and GPT Image 1 are different model families.

GPT‑4o is a general‑purpose, text-focused model that takes text and images as input and produces text output. [GPT‑4o model]
GPT Image models (gpt-image-1.5, gpt-image-1, gpt-image-1-mini) are dedicated image generation models that take text (and optionally images) as input and produce image output. [Image overview; Model comparison]

Some tools (like ChatGPT Images or the Responses API image tool) may use GPT‑4o as the main model and then call a GPT Image model under the hood, but the docs do not say that GPT‑4o is `gpt-image-1`. They are documented as separate models with different capabilities.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@latent dagger Kapa.ai is still learning and improving, please let me know how I did by reacting below.

rough rapids Apr 18, 2026, 2:47 AM

#

How can I set the image generation to gpt-image-1?

#

@narrow sparrow How can I set the image generation to gpt-image-1?

narrow sparrowBOT Apr 18, 2026, 2:48 AM

#

Let me check the knowledge base for you again. One moment! :books:

#

To use gpt-image-1 you need to select it explicitly in the Image API, or configure it in the image_generation tool (Responses API).

1. Image API (direct image generation)

Set model to "gpt-image-1" in your images.generate call:

from openai import OpenAI
import base64

client = OpenAI()

result = client.images.generate(
    model="gpt-image-1",
    prompt="A cute baby sea otter",
    size="1024x1024",
)

image_base64 = result.data[0].b64_json
image_bytes = base64.b64decode(image_base64)
with open("otter.png", "wb") as f:
    f.write(image_bytes)

The Images API explicitly supports gpt-image-1, gpt-image-1-mini, and gpt-image-1.5 as valid model values.[Images createImage]

2. Responses API (image_generation tool)

With the Responses API, you don’t put gpt-image-1 in the top‑level model field. Instead:

Use a text model (e.g. gpt-4.1, gpt-4o, gpt-5, etc.) as model
Add the image_generation tool
Optionally set the tool’s model to "gpt-image-1"

Example:

from openai import OpenAI
import base64

client = OpenAI()

response = client.responses.create(
    model="gpt-4.1-mini",
    input="Generate an image of a gray tabby cat hugging an otter with an orange scarf",
    tools=[{
        "type": "image_generation",
        "model": "gpt-image-1"
    }],
)

image_data = [
    output.result
    for output in response.output
    if output.type == "image_generation_call"
]

if image_data:
    image_base64 = image_data[0]
    with open("cat_and_otter.png", "wb") as f:
        f.write(base64.b64decode(image_base64))

#

The image_generation tool supports GPT Image models and lets you configure `model`, `size`, `quality`, etc.[Tools image-generation; Tools image-generation usage]

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@rough rapids Kapa.ai is still learning and improving, please let me know how I did by reacting below.

rough rapids Apr 18, 2026, 2:49 AM

#

@narrow sparrow Idk anything about this, so can you guide me through as to where to put it?

narrow sparrowBOT Apr 18, 2026, 2:49 AM

#

Let me check the knowledge base for you again. One moment! :books:

#

You control this in the code that calls the API. There are two main places you might set gpt-image-1:

1. Using the Image API directly

Here you call the Images endpoint and set model: "gpt-image-1" in the JSON body. [Images create]

Example with Python

from openai import OpenAI
import base64

client = OpenAI(api_key="YOUR_API_KEY")

result = client.images.generate(
    model="gpt-image-1",              # <‑‑ set it here
    prompt="A cute baby sea otter",
    size="1024x1024",
)

image_base64 = result.data[0].b64_json
image_bytes = base64.b64decode(image_base64)

with open("otter.png", "wb") as f:
    f.write(image_bytes)

Example with curl

curl <https://api.openai.com/v1/images/generations> \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-image-1",          // <‑‑ set it here
    "prompt": "A cute baby sea otter",
    "size": "1024x1024"
  }'

In both cases, the "model" field is where you put "gpt-image-1". [Generate images]

2. Using the Responses API with the image tool

Here you:

Choose a main model (like gpt-4.1-mini) in the top-level model field.
Add the image_generation tool.
Optionally set that tool’s model to "gpt-image-1". [Tools image-generation usage]

Example with Python

from openai import OpenAI
import base64

client = OpenAI(api_key="YOUR_API_KEY")

#

response = client.responses.create(
model="gpt-4.1-mini", # main chat model
input="Generate an image of a gray tabby cat hugging an otter with an orange scarf",
tools=[{
"type": "image_generation",
"model": "gpt-image-1" # <‑‑ set it here
}],
)

image_data = [
output.result
for output in response.output
if output.type == "image_generation_call"
]

if image_data:
image_base64 = image_data[0]
with open("cat_and_otter.png", "wb") as f:
f.write(base64.b64decode(image_base64))


Here, `"gpt-image-1"` goes inside the `tools` entry, not in the top-level `model`. [[Image tool options](<https://developers.openai.com/api/docs/guides/tools-image-generation#usage>)]

---

If you tell me:

- whether you’re using Python, JavaScript, or just curl, and  
- whether you want “simple: just give me an image” or “chat that can also make images”  

I can show you the exact minimal snippet you should copy‑paste and where to change only one line.
----------------------
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
----------------------

#Models for Images

In the API

1. GPT Image models (recommended)

2. DALL·E models (deprecated but still available)

In ChatGPT

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

Some tools (like ChatGPT Images or the Responses API image tool) may use GPT‑4o as the main model and then call a GPT Image model under the hood, but the docs do not say that GPT‑4o is gpt-image-1. They are documented as separate models with different capabilities.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

1. Image API (direct image generation)

2. Responses API (image_generation tool)

The image_generation tool supports GPT Image models and lets you configure model, size, quality, etc.[Tools image-generation; Tools image-generation usage]

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

1. Using the Image API directly

Example with Python

Example with curl

2. Using the Responses API with the image tool

Example with Python

Some tools (like ChatGPT Images or the Responses API image tool) may use GPT‑4o as the main model and then call a GPT Image model under the hood, but the docs do not say that GPT‑4o is `gpt-image-1`. They are documented as separate models with different capabilities.

The image_generation tool supports GPT Image models and lets you configure `model`, `size`, `quality`, etc.[Tools image-generation; Tools image-generation usage]