The launch of DALL-E 3 by OpenAI has sparked a new wave of debate within the AI community. Many wonder how OpenAI's latest image model compares to Midjourney, especially with the latter already cementing itself as one of the best AI art tools on the market. Imaigic did some research into DALL-E 3 vs. Midjourney and believes you will find the discoveries interesting. Although much of what is known about DALL-E 3 comes from information from early testers (full release comes up in October), it is enough for anyone to see where OpenAI is heading with this one. Enjoy our in-depth DALL-E 3 vs. Midjourney comparison, which begins with an overview of both tools. We also compare both AI image models across several major components, such as image generation capabilities, onboarding, pricing, image modifications, restrictions, and advanced features.
DALL-E 3 is OpenAI’s latest response to the growing competition in AI text-to-image generation models. The company markets the model as a significant improvement to its previous systems, including DALL-E 2. For instance, while DALL-E 2 noticeably struggled with understanding specific user prompts, OpenAI’s new model significantly improves translating ideas into images. The product’s demo release presents the below example of an image generated with DALL-E 2 vs. DALL-E 3 using the same prompt. 
DALL-E 2 vs. DALL-E 3 (Source: OpenAI) Beyond its enhanced understanding, another prominent feature that DALL-E 3 brings is native integration with ChatGPT. Users may decide against writing detailed text prompts and provide ChatGPT with an idea of the image they want to create. The chatbot would, in turn, feed DALL-E 3 with a clear prompt, which helps it generate the desired image output.
Midjourney launched its image-generating model in 2022 and quickly gained prominence in AI art. The tool is mainly accessible via a Discord bot, where users can type in prompts, which the AI model translates into photorealistic images. Despite being relatively new, Midjourney’s latest version, 5.0, has already demonstrated a significant ability to understand and translate user prompts. The AI image tool also offers advanced tools for refining outputs, creating variations, and editing web-based images. Although Midjourney has limitations around generating human anatomy and legible texts, the issues are general constraints facing AI image generation. Meanwhile, unlike OpenAI, which runs a startup enterprise, Midjourney is a self-funded independent research lab with a closely-knit team of 11 full-time employees. Midjourney continues to push the boundaries of AI image generation with several new feature releases for its estimated 16 million users.
This section pits DALL-E 3 against Midjourney 5.0 across different major features.
DALL-E 3 offers almost the same generative AI image capabilities as Midjourney. For this comparison, we simply prompted Midjourney 5.0 to create some images using the exact prompts OpenAI used to generate three photos found in the DALL-E 3 announcement. Prompt 1: “An expressive oil painting of a basketball player dunking, depicted as an explosion of a nebula.”
(DALL-E 3 vs. Midjourney)
Prompt 2: “A modern architectural building with large glass windows, situated on a cliff overlooking a serene ocean at sunset.”
(DALL-E 3 vs Midjourney)
Prompt 3: “A photo of an ancient shipwreck nestled on the ocean floor. Marine plants have claimed the wooden structure, and fish swim in and out of its hollow spaces. Sunken treasures and old cannons are scattered around, providing a glimpse into the past.”
(DALL-E 3 vs Midjourney)
We found that users can achieve Midjourney results that more closely match DALL-E 3 results by adding interesting tokens such as “colorful” and “hyper-realistic.” Verdict: Tie
OpenAI’s integration of DALL-E 3 with ChatGPT and Bing means that the new AI image model will probably have a seamless onboarding experience. Users can access DALL-E 3 using existing ChatGPT, Microsoft, Google, or Apple accounts. Midjourney, on the other hand, has always had a lengthy onboarding experience that requires newcomers to learn how to use Discord. Unless Midjourney delivers its widely-anticipated standalone website application in the coming months, it will likely continue to trail DALL-E 3 in user onboarding. Verdict: DALL-E 3
Early indications are that DALL-E 3 does not allow users to modify the image it initially generates directly. Although users may request the tool to rerun the provided prompt, they must use third-party AI image refiner solutions to upscale and enhance the initial results. Midjourney has an edge on this feature, allowing users to upscale or create variants. Users can simply click a button to get an upscaled or varied version of their preferred image from the initial grid of four pictures.
(Midjourney Image Modification Features)[/caption] Additionally, Midjourney provides advanced functionalities such as image blending and prompting. DALL-E 3 does not deliver these features, making Midjourney a clear winner. Verdict: Midjourney
Increased advocacy on responsible AI development means companies must adopt restrictions to keep communities civil and organized. Users are generally unable to create content deemed violent, adult-related, or hateful. However, OpenAI revealed additional eye-catching restrictions for DALL-E 3. For instance, the model will decline prompts to generate images of “public figures by name.” It will also “decline requests that ask for an image in the style of a living artist," with creators able to opt out of having OpenAI use their trademarked creations to train future DALL-E models. Midjourney does not include any of these restrictions. However, the AI model creator also maintains a range of community guidelines that allow users to responsibly harness the tool without restrictions on public figures and creator-themed artwork. Verdict: Midjourney
DALL-E 3’s native integration with ChatGPT gives it an edge over Midjourney based on this comparison metric. Users can prompt the AI using simple ideas like the one featured in the OpenAI DALL-E 3 demo below. 
Midjourney currently does not offer text generation. Even though the feature may arrive in Midjourney 6.0 and future versions, DALL-E 3 may maintain an edge, given ChatGPT’s vast training module and first-mover advantage. Verdict: DALL-E 3
The earliest version, DALL-E 3, automatically generates images with a 1024 x 1729 aspect ratio. Users also have the option to specify a square aspect ratio. Midjourney, on the other hand, offers a default resolution of 1024 x 1024, which is perfect for most artwork. Midjourney users may specify different aspect ratios and use Zoom In or Zoom Out to modify the final output. These additional features make Midjourney a preferred choice for creators who want greater control over image size. Verdict: Midjourney
OpenAI is rolling out DALL-E 3 for ChatGPT Plus and Enterprise users. The model will also be accessible for free through Bing at an undisclosed date. Barring when the Bing feature becomes available, DALL-E 3’s lowest access tier is $20, which is the monthly fee for ChatGPT Plus. In contrast, Midjourney’s lowest tier is a $10 monthly plan, which allows the generation of up to 200 images. Although Midjourney looks like a better bargain, the limited number of images and the lack of access to advanced chatbot features users enjoy with ChatGPT Plus are worth noting. Both products offer relatively competitive pricing depending on what an individual wants the most (images/chatbot). Verdict: Tie
A seed is a series of numbers that tells an AI how to generate your image. Using a seed, you can generate the same AI image on the model without worrying about getting a different picture each time you reuse a prompt. DALL-E 3 and Midjourney support using seeds for image generation, meaning there is no clear winner. Verdict: Tie
OpenAI is making DALL-E 3 accessible via API. This integration will allow users to integrate the AI image model within various applications, thus accelerating adoption. Midjourney does not offer API access, making it less suitable for third-party integrations. Verdict: DALL-E 3
The release of DALL-E 3 marks a milestone for OpenAI in its quest to grab a share of the text-to-image market. Notably, the latest model represents a significant improvement from DALL-E 2 and already has some advantages over Midjourney. For instance, DALL-E 3 provides simpler onboarding, text generation, API access, and competitive pricing. However, it trails Midjourney on other vital features, such as image resolution and modifications. Midjourney also imposes little to no restrictions on public figures and artist-specific prompts, giving users more power to explore the language model. Meanwhile, both image models offer relatively distinct features and user experiences, making them valuable for creators. The healthy competition between DALL-E 3 and Midjourney will also allow the technology to thrive and bring further sophisticated tools for users as the industry matures.
