How to get access to dall-e 3 from API


I would like to use dall-e v3 in our integration

  1. how do I do that?
  2. does if require any code changes?
  3. I am using node openai package, how do I configure it to work with dall-e 3

You read the API reference linked on the sidebar of the forum, scroll to the part about images, be familiar with the different API method for requesting, prompting, and receiving image URLs and downloading them and serving from your own site.

And then you wait for any announcement that the upgraded version would be available on the API and review the price per image and see if it makes sense…


So right now there is no way to use dall-e v3 from the API?

1 Like

If DALL-E v3 is a significant update, you may need to update your Node.js OpenAI package to a version that supports it. Check the package’s documentation or repository for any updates related to DALL-E v3 support. If there are any new configuration options specific to DALL-E v3, you’ll need to update your OpenAI package configuration to use these options. This may include specifying the model version, authentication, and other settings.

1 Like

There is no DALL-E v3 available. It’s simply not possible to use it via an API.

1 Like

Did you just paste nonsense an AI said?

No, there is currently no API access. is still DALL-E 2 for your credits.

1 Like

Damn, I think my brain mis-parsed this:

DALL·E 3 is now in research preview, and will be available to ChatGPT Plus and Enterprise customers in October, via the API and in Labs later this fall.

from: DALL·E 3

I read that originally as “October via the API and Labs later this fall”.



I’m also eagerly awaiting support for Dall E 3 via API, was hoping it would be ready early or even beginning of October.

1 Like

Nope, there should be a “then” in there for clarity.

…available to ChatGPT Plus…October, then via the API and Labs later this fall.


I was wondering the same. Looks like it’s only available through ChatGPT+ right now. All the docs are still v2. Hopefully soon!

1 Like

is it available now, would even be happy to use a limited api with like upto 100 images haha.

1 Like

No word yet, but we’ve got a lot of great dalle3 threads if you want to stick around and join our growing community.


I’m just writing so I get notified when the Dalle3 api comes out :wink:
Have been experimenting with the dalle 2 api from the start for our business but it just does not cut it enough for it to be implemented.
I’ve seen in chatgpt what it can do and those images are awesome and very usable. It takes about 25 seconds to generate 4 images ( much faster then midjourney ). If it could be set to generating just 1 image probably generation time can be cut significantly, which will make the api even more usefull.
Can’t wait for it to role out.

1 Like

Welcome. There’s quite a few of us waiting earnestly, so you’re not alone.

Feel free to stick around. We’ve got a few good dalle3 threads already and a lot of other great gems if you dig a bit. Once you hit Trust Level 3 here, you even get access to generative text AI tools to use on the forum.

1 Like

Writing here to get notified when the dall-e 3 API released


  "prompt": "A squirrel nestled in a tree, tucked in his bed, dreams of accessing the new image API. It's hard for him to sleep because he drank coffee before going to bed and the thought of DALL-E 3 is very exciting.",
  "n": 1,
  "size": "1024x1024"

(I’m also excited to hear about API options for DALL-E 3.)


Also waiting for the API! :pinched_fingers:


You don’t need to post here to be informed of new messages, just click on the bell on the bottom right of the screen :slight_smile:

I am so excited for API access to Dall E 3, hoping that "later this fall’ comes very soon!

Once released, I would like to create images with it, and then use a second prompt to review the images to confirm they contain the desired entities.

Is that possible?

You can use the images endpoint to generate DALL-E images now. They are just the previous generation, which has different qualities, such as following a prompt and making more artistic and also more realistic styles, but with a poor job on human faces and anatomy, and worse on text.

gpt-4-vision is not available by API, it is another “possibly in the future” feature. You can use other machine vision models for feature identification, although they might not have the skill to tell you if your human looks Irish.