How Many People Are Using Text-to-Image AI Like DALL-E and Midjourney?
How Many People Are Using Text-to-Image AI Like DALL-E and Midjourney?
A lot
Updated November 9, 2022
We hear a lot about text-to-image generators, not just because of the amazing artwork they produce, the impact on artists and art contests, or the numerous headlines. There are also a lot of users.
The three biggest platforms today appear to be DALL-E from OpenAI, Midjourney, and Stable Diffusion. There are many others, such as Night Café, Wombo, Craiyon, and so on. I have not yet found user data on those, but we have some indicators for the Big 3.
How Many People are Using DALL-E?
OpenAI has revealed that more than 1.5 million users have used DALL-E up through early October 2022, and they are producing more than two million images per day. Those figures were from DALL-E’s beta period. Now that it’s open to everyone, figures have risen quickly. As of November 2022, OpenAI says more than three million people are using DALL-E, and they are creating four million images per day. That is a doubling of the user base and image generation volume in just a month.
You should expect OpenAI to announce in the next six months that more than one billion images have been created by platform users.
How Many People are Using Midjourney?
Midjourney CEO and co-founder David Holz said in a September interview that millions are using Midjourney and that more than two million people are members of its Discord community. For Midjourney, Discord is where users interface with the text-to-image AI model to generate novel artwork. Holz also said he believes more than 30% and maybe as high as 50% of the users are art professionals.
The Midjourney web app is still in its beta period. So, most of its users appear to be accessing the model through Discord. So, the figure for registered users is likely between 2-3 million.
How Many People are Using Stable Diffusion?
This may shock you. Bloomberg reported that Stability AI already had more than 10 million users. And Stability AI is the creator of the open source model but is not the only provider. There are likely many more Stable Diffusion users on private servers and through other text-to-image application providers. If these figures are true, it also means that Stable Diffusion was more than two times more users than DALL-E and Midjourney combined as of October 2022.
However, updated numbers for DALL-E in November 2022 show rapid growth. The new API for DALL-E will only increase these figures. Granted, Stable Diffusion will likely grow even faster, given it has both demand growth from Stability AI and the open source community.
The Tally
The public tally was more than 13.5 million registered users across the Big 3 platforms combined as of October 2022. However, there was surely some double-counting as many users have tried multiple platforms. Some quick math that takes a haircut for the multi-platform users and adds in some unique users from the platforms beyond the Big 3 suggests that total registered users were likely in the range of 8-11 million as of October 2022. Given the new numbers from OpenAI around DALL-E users, that figure will be 9.5 - 12.5 million as of early November 2022.
However, I assume this does not include users of the DALL-E API, such as Simplified and Jasper AI. These could easily add a few million casual users even at this early stage. We will be doing some more modeling and making updates to this article in the future.
This figure will surely grow over the next three months. Stability AI just raised $101 million, and Jasper AI, which is now offering access to DALL-E, raised $125 million. OpenAI is flirting with a new funding round and a valuation of $20 billion. All of these companies will be focused on driving new user growth.
Granted, many of these registered users experimented and then departed. User retention will be an important metric in gauging the appeal and staying power of text-to-image generators. However, there is little argument that the segment has emerged with a fast start.
