Midjourney

The world's leading AI image generation platform generates high-quality artistic images in about 60 seconds through text prompts

image generation

https://www.midjourney.com/

Visit Website

In-depth Report

Midjourney is the world's leading AI image generation platform, founded by David Holz in 2021 in San Francisco, USA. As an independent research laboratory, Midjourney operates on a self-funded model and provides Vincentian graphics services through the Discord platform. Users only need to enter a short text prompt to generate high-quality artistic images in about 60 seconds. With its unique artistic style and active Discord community ecology, Midjourney has become a phenomenal application in the AIGC field, achieving more than 10 million users and approximately US$100 million in revenue within one year. The V7 version released in April 2025 made a breakthrough in text rendering, and the V8.1 Alpha version launched in April 2026 increased the rendering speed by 5 times while achieving native 2K high-definition output.
Midjourney was founded in August 2021 by former Leap Motion founder David Holz and is headquartered in San Francisco, USA. The team has grown from an initial 11 people (eight of whom are undergraduates) to approximately 40 people by the end of 2023. As an independent research lab, Midjourney adheres to a self-funded model, never accepting outside venture capital, and became profitable six months after launch. The core concept of Midjourney is that "AI is not a replica of the real world, but an extension of human imagination." The product runs on the Discord platform. Users can quickly generate images by talking to the Midjourney bot and inputting text prompts. This unique interaction model not only lowers the threshold for use, but also accumulates a large amount of high-quality "text-image" annotation data through community interaction, forming a significant data flywheel effect.
The underlying technology of Midjourney is based on the Diffusion model, which generates images by adding noise to the training data and then "reversely extending" the recovered data. This technical route is simple to interact with and fast to generate. It only takes about 60 seconds for users to generate a picture. The core function upgrades of the V7 version include: a breakthrough in text rendering capabilities, which can accurately generate English letters and simple Chinese, and can also better control font size and font style; personalization settings are turned on by default, and new users only need to evaluate about 200 pictures to establish a personal preference profile; Draft Mode generates images at 10 times faster and 50% of the cost, suitable for quick brainstorming; the voice interaction function supports voice input commands; the natural language editor supports directly describing editing needs in natural language. The V8.1 Alpha version further achieves 5x faster image rendering, coupled with native 2K high-definition generation, the artistic sense is still top-notch. The new `--seed` and `--style reference` parameters can maintain the style consistency of multiple generations of images. The Region Vary tool supports precise editing of parts of the image without regenerating the entire image. Control of faces, hands and complex scenes has also been significantly improved.
Midjourney uses a B2C subscription model with tiered pricing based on GPU usage and feature access. All plans are automatically renewed and can be paid monthly or annually. If you choose annual payment, you can enjoy about 20% off. In terms of annual recurring revenue, Midjourney's revenue in 2023 will be approximately US$200 million (approximately US$114 million), with more than 21 million Discord community members, approximately 1.4 million paid subscribers, and an average monthly subscription price of approximately US$30. The specific subscription plans are divided into four levels: the Basic plan costs US$10 per month (annual payment is US$96, average monthly payment is US$8), providing 3.3 hours of Fast GPU time, suitable for casual users; the Standard plan costs US$30 per month (annual payment is US$288, average monthly payment is US$24), provides 15 hours of Fast GPU time and unlimited Relax mode, and is suitable for advanced creators; the Pro plan costs US$60 per month (annual payment is US$576) USD, $48 monthly average), offers 30 hours of Fast GPU time, unlimited images and videos, Stealth privacy mode, suitable for professional users; the Mega plan costs $120 monthly ($1,152 annually, average monthly $96), provides 60 hours of Fast GPU time, suitable for heavy users and studios.
According to user review data on Product Hunt, Midjourney has an overall score of 4.8/5.0, based on 80 reviews. Positive user feedback mainly focused on: high-quality image generation (20 mentions), AI-driven image generation (15 mentions), efficient creative exploration (12 mentions), user-friendly interface (8 mentions), powerful AI algorithm (5 mentions), and unique artistic style (5 mentions). Negative user feedback focused on: limited control over final image (7 mentions), expensive pricing (6 mentions), Discord integration issues (5 mentions), limited free trial (4 mentions), content filtering issues (3 mentions). In terms of professional evaluation, TechCrunch reporter Kyle Wiggers believed that the V7 model's compliance with prompt words was "quite reasonable" after testing, and the sketch test was basically satisfactory. Decrypt reviews pointed out that V7 is an "incremental upgrade" rather than a revolutionary update. Competitors GPT-4o and Reve have surpassed Midjourney in terms of realism and accuracy. CogitoDaily's in-depth experience analysis believes that V8 is "an intergenerational leap", with facial control and complex scene processing capabilities significantly improved, and it is still the "king of aesthetics" in the field of AI art.
In terms of industry impact, Midjourney has had a profound impact on the creative industries. In terms of positive impact, it has successfully promoted the implementation of AIGC applications on the C side, allowing ordinary users to become "designers"; it has changed the design process, stimulates creativity and improves efficiency; it may replace traditional photo galleries such as Google Images and Shutterstock. In terms of potential challenges, copyright issues are highly controversial, and training data containing unauthorized works has triggered multiple lawsuits; facing regulatory risks, the compliance of AI-generated content requires attention; technology iteration is rapid, and new technologies may subvert the current product form; as an upper-level application, there is a lack of solid technical barriers. From the competitive landscape, the main competitors include: DALL-E 3 (OpenAI) - the strongest text understanding ability, deeply integrated with ChatGPT; Stable Diffusion - open source model, completely free and can be deployed locally; Adobe Firefly - integrated design software ecosystem; domestic models such as Jimeng AI and Wen Xinyiyan - localized services and deep penetration of industry scenarios.
Midjourney is currently facing multiple copyright lawsuits. In June 2025, Disney and Universal Pictures jointly sued Midjourney, accusing it of grabbing copyrighted works without authorization to train AI models. In September 2025, Warner Bros. also filed a copyright lawsuit, accusing Midjourney of allowing users to generate images of characters such as Superman and Batman. The plaintiff's core claims include: Midjourney is a "bottomless abyss of plagiarism" that generates "endless unauthorized copies"; the output is "not transformative enough" and the fair use defense is difficult to establish; AI-generated art is a broader threat to the American film industry, which contributes more than $260 billion to the U.S. economy. Midjourney’s defense includes: training AI requires the use of “billions of publicly available images”; training extracts statistical information from copyrighted works, which is a “transformative fair use”; citing author copyright lawsuit judgments against Anthropic and Meta as support.
As a phenomenal application in the field of AIGC, Midjourney has achieved commercial success by virtue of the following advantages: stunning image generation effects and significant differentiation in artistic styles; strong profitability, with annual revenue of approximately US$200 million; community advantages and data flywheels forming a certain moat; the team is streamlined and efficient, and maintains strong independence. Challenges faced include: the risk of copyright litigation continues to intensify; the lack of solid technical barriers as an upper-level application; competition intensifies, with many domestic and foreign pursuers; rapid technological iteration, requiring continuous innovation. Core conclusion: Midjourney is one of the most competitive applications in the field of Vincentian graphics. It can maintain strong profitability in the short term but faces the risk of technological subversion in the long term. It is recommended that creative design practitioners use Midjourney as a productivity tool, but need to pay attention to copyright compliance and alternatives.

User Reviews

MsPietroMenger_x

—

Let me talk about a money-saving tip that many people don’t know: the Relax Mode of the Standard Plan ($30/month) actually has unlimited drawings, but you may have to wait 2-5 minutes during peak periods. I had been using Fast Mode stupidly before, and the 900 cards a month were quickly used up. Later, I changed to using Relax to iterate the direction first, and then cut to Fast to produce the final picture. I can actually produce 2,000+ pictures in a month, which is equivalent to a cost of less than one and a half cents per picture. But this technique is only for individual creators who are not in a hurry to publish pictures. If you can’t afford to wait for commercial projects, you should just stick to Pro.

BrianWard_7

—

Let me talk about a money-saving tip that many people don’t know: the Relax Mode of the Standard Plan ($30/month) actually has unlimited drawings, but you may have to wait 2-5 minutes during peak periods. I had been using Fast Mode stupidly before, and the 900 cards a month were quickly used up. Later, I changed to using Relax to iterate the direction first, and then cut to Fast to produce the final picture. I can actually produce 2,000+ pictures in a month, which is equivalent to a cost of less than one and a half cents per picture. But this technique is only for individual creators who are not in a hurry to publish pictures. If you can’t afford to wait for commercial projects, you should just stick to Pro.

SSullivan

—

MJ's recent lawsuits with several major Hollywood studios are quite interesting. Disney and Warner Bros. sued MJ over its ability to generate copyrighted character images, and MJ in turn demanded that the studio disclose that they were also using AI internally for storyboarding and concept design. If this case is decided, it may directly affect the direction of the AI painting industry - what is "fair use" and what is "infringement". For those of us who make content, using MJ for commercial projects is actually a gray area now. Although the platform says it includes commercial use rights, the compliance of the training data is still questionable. It is recommended that commercial users be careful and at least not rely entirely on MJ's output.

ur_bzy

—

MJ's recent lawsuits with several major Hollywood studios are quite interesting. Disney and Warner Bros. sued MJ over its ability to generate copyrighted character images, and MJ in turn demanded that the studio disclose that they were also using AI internally for storyboarding and concept design. If this case is decided, it may directly affect the direction of the AI painting industry - what is "fair use" and what is "infringement". For those of us who make content, using MJ for commercial projects is actually a gray area now. Although the platform says it includes commercial use rights, the compliance of the training data is still questionable. It is recommended that commercial users be careful and at least not rely entirely on MJ's output.

Mason.Wood36905

—

MJ's community ecosystem is really good. There are 20M users in the official Discord, and there are new prompts to share and learn from every day. For novices, it is much more efficient to see how others write prompts than to try blindly by yourself. This is one of the reasons why I stayed with MJ after comparing other tools.

JChavez007113

—

MJ's community ecosystem is really good. There are 20M users in the official Discord, and there are new prompts to share and learn from every day. For novices, it is much more efficient to see how others write prompts than to try blindly by yourself. This is one of the reasons why I stayed with MJ after comparing other tools.

ANrob

—

I do character concept design for independent games. After the MJ V7's personalized system has been adjusted, the style consistency of the pictures is much more stable than before. In the past, parameters had to be adjusted repeatedly for each picture, but now they are basically formed in one go. The disadvantage is that there is no LoRA, so you have to rely on sref hard adjustment to change the style.

PMiller3

—

I do character concept design for independent games. After the MJ V7's personalized system has been adjusted, the style consistency of the pictures is much more stable than before. In the past, parameters had to be adjusted repeatedly for each picture, but now they are basically formed in one go. The disadvantage is that there is no LoRA, so you have to rely on sref hard adjustment to change the style.

Noah.Russell_X217

—

Midjourney is really easy to use for creating social media materials. As long as you don’t write words in the pictures, it’s basically a one-time use.

DanielDavis_77

—

Midjourney is really easy to use for creating social media materials. As long as you don’t write words in the pictures, it’s basically a one-time use.

NAsmi

—

Ji Meng is getting stronger and stronger, and the Chinese understanding is really incomparable to MJ.

Lisa.Campbell_Plus768

—

Ji Meng is getting stronger and stronger, and the Chinese understanding is really incomparable to MJ.

muab2xts3

—

Stealth mode is only available on Pro, and it costs $60 a month just for privacy, which is really too much for a small team.

Scott.Fisher_Max36

—

Stealth mode is only available on Pro, and it costs $60 a month just for privacy, which is really too much for a small team.

Gabriel.Stewart007

—

After Flux came out, the gap is actually not that big, but MJ's aesthetic is still unique.

JeffreyWeber_pro

—

After Flux came out, the gap is actually not that big, but MJ's aesthetic is still unique.

AColeman168

—

It took two years for Discord to finally be released on the Web. There is no need to switch to the chat box to enter commands, and the experience has been greatly improved.

AmandaGreen_2021

—

It took two years for Discord to finally be released on the Web. There is no need to switch to the chat box to enter commands, and the experience has been greatly improved.

DHughes_66

—

As an interior designer, I use MJ to demonstrate design directions to clients. It used to take a week to find reference drawings for three sets of plans, but now I can come up with more than a dozen styles in one morning. But there is a big problem: the spaces generated by MJ often have irrational physical structures - such as suspended stairs and illogical load-bearing walls. It’s okay to show the concept direction to the client, but if you really want to put it into practice, you still have to re-draw the CAD by yourself. It is recommended that colleagues only use it as an inspiration tool and not directly as a construction drawing.

Emily.RiveraZ

—

As an interior designer, I use MJ to demonstrate design directions to clients. It used to take a week to find reference drawings for three sets of plans, but now I can come up with more than a dozen styles in one morning. But there is a big problem: the spaces generated by MJ often have irrational physical structures - such as suspended stairs and illogical load-bearing walls. It’s okay to show the concept direction to the client, but if you really want to put it into practice, you still have to re-draw the CAD by yourself. It is recommended that colleagues only use it as an inspiration tool and not directly as a construction drawing.

Alice_RossQ

—

After using it for half a year, I have to say that you get what you pay for, which is much better than the free one.

xPárisBarros_2024

—

After using it for half a year, I have to say that you get what you pay for, which is much better than the free one.

Nicholas.Powell52044

—

The character consistency function of V7 can finally be used. Cref cooperates with sref, so the same character will not always change his face in different scenes.

Heather.Price_775

—

The character consistency function of V7 can finally be used. Cref cooperates with sref, so the same character will not always change his face in different scenes.

Rebecca.Gomez_Max6

—

The Relax mode is really nice. I can hang a bunch of pictures at night and wake up the next day, which means unlimited pictures.

SamanthaMurray42

—

The Relax mode is really nice. I can hang a bunch of pictures at night and wake up the next day, which means unlimited pictures.

PamelaRivera_99

—

V8.1’s hand finally stopped falling apart! Happy celebration!

RSanchez

—

V8.1’s hand finally stopped falling apart! Happy celebration!

武浩

—

I have compared three tools: MJ, FLUX, and SD3.5. MJ is indeed the strongest in terms of the aesthetic quality of a single image, FLUX is stronger in realism and text, and SD3.5 is stronger in controllability. When doing commercial projects, I usually use a mixture of methods: MJ gives direction first, and then SD refines the details. But the cost is also the highest.

ZetaZoneLarsen

—

I have compared three tools: MJ, FLUX, and SD3.5. MJ is indeed the strongest in terms of the aesthetic quality of a single image, FLUX is stronger in realism and text, and SD3.5 is stronger in controllability. When doing commercial projects, I usually use a mixture of methods: MJ gives direction first, and then SD refines the details. But the cost is also the highest.

Midjourney

In-depth Report

User Reviews

Similar Products