I changed the configuration file to DALL-E 3, but the final drawing API still calls DALL-E 2, which is strange.

[mithew]

I changed the configuration file to DALL-E 3, but the final drawing API still calls DALL-E 2, which is strange.

[TerminalTorture]

I think its under open_utils.py

Since no model is listed it defaults to dalle-2

async def generate_images(prompt, n_images=4, size="512x512"): r = await openai.Image.acreate(prompt=prompt, n=n_images, size=size) image_urls = [item.url for item in r.data] return image_urls

change it to

async def generate_images(prompt, n_images=1, size="1024x1024"):
r = await openai.Image.acreate(model="dall-e-3", prompt=prompt, n=1, size=1024x1024)
image_urls = [item.url for item in r.data]
return image_urls

Here is the link to the API example https://platform.openai.com/docs/guides/images/usage?context=node Do note that there are some defaults

[JonahWhaler]

I changed the configuration file to DALL-E 3, but the final drawing API still calls DALL-E 2, which is strange.

@mithew Can you please let us know your openai version? It's set at requirements.txt. @TerminalTorture , I believe the site you attached is based on the latest version which is 1.3.5, but the version used in this repo is 0.28.1.

# bot/openai_util.py
async def generate_images(prompt, n_images=4, size="512x512"): 
  r = await openai.Image.acreate(prompt=prompt, n=n_images, size=size)
  image_urls = [item.url for item in r.data]
  return image_urls

# openai/api_resources/image.py/Image
@classmethod
async def acreate(
  cls,
  api_key=None,
  api_base=None,
  api_type=None,
  api_version=None,
  organization=None,
  **params
):
  # bla bla bla

[TerminalTorture]

async def generate_images(prompt, n_images=1, size="1024x1024"):
    r = await openai.Image.acreate(model="dall-e-3", prompt=prompt, n=n_images, size=size)
    image_urls = [item.url for item in r.data]
    return image_urls

Running on version Open AI 0.28.1

[JonahWhaler]

async def generate_images(prompt, n_images=1, size="1024x1024"):
    r = await openai.Image.acreate(model="dall-e-3", prompt=prompt, n=n_images, size=size)
    image_urls = [item.url for item in r.data]
    return image_urls

Running on version Open AI 0.28.1

Cool. It works for me too. Thanks!

[mithew]

async def generate_images(prompt, n_images=1, size="1024x1024"):
    r = await openai.Image.acreate(model="dall-e-3", prompt=prompt, n=n_images, size=size)
    image_urls = [item.url for item in r.data]
    return image_urls

Running on version Open AI 0.28.1

Success, thank you very much

[mithew]

async def generate_images(prompt, n_images=1, size="1024x1024"):
    r = await openai.Image.acreate(model="dall-e-3", prompt=prompt, n=n_images, size=size)
    image_urls = [item.url for item in r.data]
    return image_urls

Running on version Open AI 0.28.1

😄Do you know how to write code that can integrate this project with the OpenAI GPT-4 vision API?

[JonahWhaler]

async def generate_images(prompt, n_images=1, size="1024x1024"):
    r = await openai.Image.acreate(model="dall-e-3", prompt=prompt, n=n_images, size=size)
    image_urls = [item.url for item in r.data]
    return image_urls

Running on version Open AI 0.28.1

😄Do you know how to write code that can integrate this project with the OpenAI GPT-4 vision API?

Why don't you post it as a new issue. As far as I am aware, currently gpt-4-1106-vision-preview is not available at Playground.