The results created by openAI api DALLE are extremely bad

I tried to generate images with DALL-E using the same prompts in Copilot and with the DALL-E model in the OpenAI framework on iOS.
The results generated by Copilot are awesome, but the results I receive from the OpenAI API are so bad.
Does the API use a different model?
Is there any way to fix this?

On the left are results generated by copilot and on the right are results generated by OpenAI API.

Here are the prompts and the simple code.

     let prompt1 = "Design an abstract logo of letter M from colored puzzles on white background PNG format."
     let prompt2 = "Design a sleek, modern logo that integrates the letter 'B' with the silhouette of a wolf's head, positioned to use the negative space within the 'B' effectively. The logo should be monochromatic, using shades of black and gray to create contrast, set against a pure white background suitable for a PNG format."
import OpenAI

         let openAI = OpenAI(apiToken:"some-token")

         let query = ImagesQuery(prompt: prompt1,
                                 n: 4,
                                 size: "1024x1024",
                                 quality: "hd")

        let result = try await openAI.images(query: query)

I read the same thing, so I’m not even trying the API approach.
From what I’ve seen, it might be related to ‘vivid’ and ‘natural’ settings. Take a look at the other posts here tagged with dalle3, and please let me know if you’ve found a solution.