CtrlK
Midjourney is an artificial intelligence image generation program and independent research lab headquartered in San Francisco, California. Founded in August 2021 by David Holz, the platform uses diffusion models to generate images from natural language text descriptions known as prompts. Since its open beta launch on July 12, 2022, Midjourney has grown into one of the most widely used generative AI tools in the world, attracting over 20 million registered users by the end of 2024 and generating an estimated $500 million in annual revenue by mid-2025. The company operates without venture capital funding and has remained profitable since its second month of operation.
Midjourney, Inc. describes itself as "an independent research lab exploring new mediums of thought and expanding the imaginative powers of the human species." The company was founded by David Holz, who previously co-founded Leap Motion, a sensor technology company that developed hand-tracking devices for computer interaction. Leap Motion was later acquired by British firm Ultrahaptics (now Ultraleap) in 2019. After departing from Leap Motion in 2021, Holz assembled a small team of engineers to build what would become Midjourney.
The company has maintained a remarkably lean structure throughout its rapid growth. When it launched in 2022, the team consisted of roughly 11 people. By 2025, the workforce had grown to approximately 130 to 160 employees, depending on the source. This small headcount relative to the company's revenue makes Midjourney one of the most efficient AI companies in the industry, generating over $5 million per employee annually.
Notably, Midjourney has never raised external venture capital. David Holz has repeatedly declined investment offers, choosing instead to bootstrap the company entirely through subscription revenue. The company reached profitability in August 2022, just one month after launching its open beta. Revenue grew from an estimated $50 million in 2022 to $200 million in 2023 and approximately $300 million in 2024, reaching $500 million by 2025. Holz achieved this growth with $0 spent on traditional marketing, relying instead on organic community growth and word of mouth.
In 2025, TIME magazine named David Holz to its list of the 100 Most Influential People in AI.
David Holz began working on the Midjourney project in 2021, motivated by the rapid advances in AI image generation following OpenAI's release of CLIP (Contrastive Language-Image Pre-training). CLIP demonstrated that AI systems could effectively assess the alignment between text descriptions and generated images, sparking a wave of development in text-to-image generation. Holz set up an independent lab in San Francisco with an initial team of 10 engineers to develop the platform.
Midjourney launched a Discord server in February 2022, which served as the primary interface for early users. The first model version (V1) was released that same month, producing abstract and painterly images with limited coherence. An invitation-only private beta launched around March 2022, allowing a select group of users to test the system. On July 12, 2022, the open beta was announced, making the tool available to the general public through Discord.
The open beta attracted users at a remarkable pace. By the end of 2022, Midjourney had released its fourth model version and was rapidly iterating on image quality. The platform's social nature, where all generated images were visible in public Discord channels, created a built-in viral loop. Users could see what others were creating, learn from their prompts, and share striking results across social media.
In September 2022, the platform received significant mainstream attention when Jason Allen won first place in the digital arts category at the Colorado State Fair with an image called "Theatre D'opera Spatial" that he had created using Midjourney. The event ignited widespread debate about AI-generated art and its place in creative competitions.
In March 2023, a Midjourney-generated image of Pope Francis wearing a white Balenciaga-style puffer jacket went viral on social media, fooling millions of people who believed it was a real photograph. The incident, described as "the first real mass-level AI misinformation case," prompted Midjourney to end its free trial program shortly afterward.
Throughout 2023 and 2024, Midjourney continued to improve its models and expand beyond Discord. In 2024, the company launched an independent web interface at midjourney.com, allowing users to generate images without needing a Discord account. The web interface has been iteratively improved and now serves as a primary access point alongside the Discord bot.
In June 2025, Midjourney released its first video generation model, marking its entry into the AI video space. In August 2025, Meta announced a partnership with Midjourney to license its image and video generation technology for use across Facebook, Instagram, and WhatsApp.
Midjourney uses closed-source, proprietary software based on latent diffusion model architecture. While the company does not publicly disclose its exact model architecture or training data, the general approach is well understood in the research community. The system combines a large language model for text encoding with a diffusion model for image generation.
The image generation process involves several key stages:
Text encoding: The user's text prompt is processed by a text encoder, likely a variant of CLIP or a similar transformer-based model. This converts the natural language prompt into dense vector embeddings that capture semantic meaning, artistic style, color, mood, and compositional intent.
Latent diffusion: The text embeddings guide a diffusion process in a compressed latent space. Starting from random Gaussian noise, the model iteratively denoises the latent representation over many steps. A U-Net architecture with cross-attention mechanisms aligns the denoising process with the text embeddings at each step, ensuring the generated image matches the prompt's intent.
Decoding: A Variational Autoencoder (VAE) decoder converts the final latent representation back into pixel space, producing the output image.
Midjourney's models are trained on a large dataset of images sourced from the internet, including images from LAION's open-source dataset. The company has also incorporated user feedback data into its training pipeline, using information about which images users upscale, favorite, or select as a form of reinforcement learning signal to improve aesthetic quality.
Midjourney provides two primary interfaces for generating images:
| Interface | Description | Availability |
|---|---|---|
| Discord Bot | The original interface, accessible through Midjourney's official Discord server or any Discord server where the bot has been added. Users type commands in chat channels to generate images. | Available since February 2022 |
| Web Application | A standalone web interface at midjourney.com with a visual prompt builder, image gallery, settings panel, and conversation mode for iterative refinement. | Launched in 2024; available to all subscribers |
On Discord, users interact with the Midjourney Bot by typing slash commands in text channels. The bot processes the request and returns results directly in the channel. On the web interface, users type prompts into an "Imagine" bar and can adjust settings such as aspect ratio, model version, stylization, and speed through a sidebar panel.
Midjourney offers a range of commands and features for creating and modifying images. The core workflow involves generating an initial set of images, then refining the results through upscaling, variations, and other editing tools.
| Command | Description |
|---|---|
/imagine | The primary command for generating images. Users provide a text prompt, and Midjourney returns a grid of four image previews. |
/blend | Uploads and combines two to five images, merging their concepts, moods, and visual elements into a single blended output. |
/describe | Accepts an uploaded image and returns four text prompts that could reproduce a similar image, useful for reverse-engineering visual styles. |
/shorten | Analyzes a long prompt and identifies the most impactful tokens, suggesting shorter versions that preserve the core meaning. |
/settings | Opens a panel to configure default parameters such as model version, stylization level, and generation mode. |
/subscribe | Directs the user to the subscription management page. |
After generating an initial grid of four images, users can refine their results using a set of modification tools:
| Tool | Description |
|---|---|
| Upscale (Subtle) | Increases the resolution of a selected image while preserving its original details and style closely. |
| Upscale (Creative) | Increases resolution while adding new details, potentially introducing fresh visual elements. |
| Variations (Subtle) | Generates new versions of a selected image with small, detailed changes that stay close to the original composition. |
| Variations (Strong) | Generates new versions with more significant changes while maintaining the overall theme. |
| Zoom Out | Expands the canvas outward in all directions, generating new content around the original image at 1.5x or 2x zoom levels (outpainting). |
| Pan | Extends the image in a specific direction (up, down, left, or right) while keeping the original content in place. |
| Vary Region | Allows users to select a specific region of the image and regenerate only that area, functioning as an inpainting tool. |
The web-based editor, introduced alongside V7, provides a layered editing environment where users can make targeted modifications to generated images. The editor supports conversational interaction, allowing users to describe changes in natural language (for example, "replace the cat with an owl" or "change the sky to sunset").
Midjourney supports a variety of parameters that users can append to their prompts to control the output. Effective use of these parameters is a key part of prompt engineering for visual generation.
| Parameter | Syntax | Range | Description |
|---|---|---|---|
| Aspect Ratio | --ar | e.g., 16:9, 3:2 | Sets the width-to-height ratio of the generated image. The default is 1:1 (square). |
| Stylize | --s | 0 to 1000 | Controls how much creative freedom Midjourney has over the prompt. Lower values produce images that match the prompt more literally; higher values favor aesthetic quality and artistic interpretation. Default is 100. |
| Chaos / Variety | --c | 0 to 100 | Determines how varied the four images in each grid are relative to one another. Higher values push the outputs in more divergent, unpredictable directions. |
| Weird | --w | 0 to 3000 | Introduces unusual, unconventional, or quirky aesthetics. Higher values produce increasingly strange and experimental results. |
| Quality | --q | 0.25, 0.5, 1, 2 | Controls the amount of compute time spent on generation. Higher values increase detail but use more GPU time. |
| Version | --v | e.g., 5.2, 6, 7 | Specifies which model version to use for generation. |
| Niji | --niji | e.g., 5, 6, 7 | Switches to the Niji model, optimized for anime and illustrative styles. |
| Style | --style | raw, cute, expressive, scenic | Adjusts the rendering style. The raw style reduces Midjourney's default aesthetic biases. |
| No | --no | text | Negative prompting: specifies elements that should be excluded from the image. |
| Seed | --seed | 0 to 4294967295 | Sets a specific seed number for reproducibility. Using the same seed and prompt produces similar results. |
| Tile | --tile | (flag) | Generates images that can tile seamlessly, useful for patterns and textures. |
| HD | --hd | (flag) | Generates native 2K resolution images (introduced in V8 Alpha). |
Effective Midjourney prompts tend to follow several general principles:
--no red parameter.Midjourney has released multiple model versions since its initial launch, each representing significant improvements in image quality, prompt understanding, coherence, and speed. The following table summarizes all major versions.
| Version | Release Date | Key Improvements |
|---|---|---|
| V1 | February 2022 | First model release. Produced abstract, painterly images with limited coherence. Demonstrated basic text-to-image capability. |
| V2 | April 2022 | Improved coherence and visual clarity over V1, though outputs still had a distinctly abstract quality. |
| V3 | July 2022 | Introduced the --stylize and --quality parameters. Better handling of complex scenes and improved overall image fidelity. |
| V4 | November 2022 | Major quality leap. Dramatically improved detail, composition, and realism. Widely considered the version that brought Midjourney into mainstream attention. Introduced the initial Niji model for anime-style generation. |
| V5 | March 2023 | Pushed toward photographic realism with higher resolution output. More accurate prompt interpretation, broader stylistic range, and improved hand rendering. |
| V5.1 | May 2023 | Refinement of V5 with stronger default aesthetics and improved coherence. Introduced the --style raw parameter for less opinionated outputs. |
| V5.2 | June 2023 | Further aesthetic improvements, better prompt understanding, and introduction of the Zoom Out (outpainting) and /shorten features. |
| V6 | December 2023 | Enhanced accuracy for longer and more complex prompts. Improved coherence, world knowledge, and the ability to render text within images. Advanced image prompting and remixing capabilities. |
| V6.1 | July 2024 | Refinement of V6 with more coherent images, more precise details and textures, and approximately 25% faster generation speed. |
| V7 | April 2025 (alpha); June 2025 (default) | Complete rebuild from scratch. Introduced Draft Mode (10x faster, 50% cost). Significantly better body, hand, and object coherence. Default model personalization. Conversational editing mode on web interface. |
| V8 Alpha | March 17, 2026 | Fastest model to date, with standard renders 4 to 5 times faster than previous versions. Native 2K resolution via --hd parameter. Improved prompt adherence and text rendering accuracy. Currently available only on alpha.midjourney.com. |
Alongside its main model line, Midjourney offers the Niji series, developed in collaboration with Spellbrush, which specializes in anime, manga, and illustrative styles influenced by Eastern aesthetics.
| Niji Version | Release Date | Key Features |
|---|---|---|
| Niji 4 | December 2022 | First Niji model, designed for anime-style outputs. |
| Niji 5 | April 2023 | Added --style cute, --style expressive, and --style scenic sub-styles. |
| Niji 6 | June 2024 | Improved Japanese and Chinese text rendering. Greater detail and consistency. |
| Niji 7 | January 2026 | Major boost in coherency. Cleaner, flatter look with improved line work. Better prompt adherence for specific character designs. |
Midjourney operates on a subscription-based model. As of 2025, the platform no longer offers any free access, having discontinued its free trial in April 2023 following the viral Pope Francis deepfake incident. All plans include commercial usage rights and access to the member gallery.
| Plan | Monthly Price | Annual Price (per month) | Fast GPU Hours | Relax Mode | Stealth Mode | Max Concurrent Fast Jobs |
|---|---|---|---|---|---|---|
| Basic | $10 | ~$8 | ~3.3 hours | No | No | 3 |
| Standard | $30 | ~$24 | ~15 hours | Unlimited | No | 3 |
| Pro | $60 | ~$48 | ~30 hours | Unlimited | Yes | 12 |
| Mega | $120 | ~$96 | ~60 hours | Unlimited | Yes | 12 |
Fast GPU time is used for priority processing and provides the fastest generation speeds. Once fast hours are depleted, users on the Standard plan and above can switch to Relax Mode, which queues jobs at lower priority with variable wait times. Unused fast hours do not roll over between billing cycles.
Stealth Mode, available only on Pro and Mega plans, prevents generated images from appearing in public galleries on the Midjourney website. This feature is important for commercial users who need to keep their creative work confidential.
Additionally, Turbo Mode is available for all subscribers, generating images at an even faster speed at the cost of approximately double the GPU time.
On June 19, 2025, Midjourney released its first video generation model, designated V1 Video. The feature uses an image-to-video workflow: users first generate a still image through the standard image generation process, then press an "Animate" button to convert it into a short video clip.
Key details of the video generation system include:
Midjourney operates in an increasingly competitive AI image generation market. While each platform has distinct strengths, Midjourney has consistently been recognized for its strong aesthetic quality and artistic output.
| Platform | Developer | Approach | Key Strengths |
|---|---|---|---|
| Midjourney | Midjourney, Inc. | Closed-source, subscription-based | Artistic quality, rich visual aesthetics, cinematic and painterly output. Strong community. |
| DALL-E 3 | OpenAI | Closed-source, API and ChatGPT integration | Precise prompt following, beginner-friendly interface, tight integration with ChatGPT for conversational image generation. |
| Stable Diffusion | Stability AI | Open-source, locally runnable | Full customization, fine-tuning capability, local processing for privacy. Free to use. |
| Flux | Black Forest Labs | Open-source and commercial tiers | Photorealistic output, strong detail and lighting. Flux Pro produces some of the most realistic AI images available. |
| Adobe Firefly | Adobe | Integrated into Creative Cloud | Commercially safe (trained on licensed content), seamless integration with Photoshop, Illustrator, and other Adobe tools. |
| Imagen | Google DeepMind | Closed-source, integrated into Google products | Strong photorealism and text rendering, available through Google's Vertex AI and consumer products. |
Midjourney's competitive advantage lies primarily in its aesthetic quality. The platform consistently produces images with a distinctive cinematic, richly detailed look that many users and professionals find superior to alternatives for artistic and conceptual work. However, for tasks requiring precise control over specific elements, exact text rendering, or photorealistic product photography, some competitors may perform better.
Midjourney has had a significant impact on the creative landscape since its launch. The platform has been adopted by a wide range of users, from amateur hobbyists exploring their imagination to professional architects, graphic designers, concept artists, and advertising agencies using it to rapidly prototype visual ideas.
Architects and interior designers use Midjourney to quickly visualize design concepts before committing to detailed plans. Graphic designers use it to generate initial mood boards and explore creative directions. Game studios and film production companies have used it for concept art during pre-production. Advertising agencies have adopted the tool for rapid ideation, generating dozens of visual concepts in the time it would traditionally take to produce a single sketch.
Midjourney's Discord-based community has fostered a culture of collaborative experimentation. Public channels function as a live gallery where users can observe each other's prompts and results, learn techniques, and share discoveries. This social dimension has been a key factor in the platform's growth and in the broader popularization of AI-generated art.
The platform has also been used in music and entertainment. Hungarian DJ and animator Shane 54 used Midjourney to generate every visual element for a music video, creating a cohesive animated collage entirely from AI-generated imagery. Several magazines, including The Atlantic and The Economist, have used AI-generated artwork for their covers.
The rise of Midjourney and similar tools has sparked significant debate within the art community. Some artists have embraced the technology as a powerful new creative tool. Others have expressed deep concern about its potential to reduce demand for human illustrators and designers. Cartoonist Matt Borrs captured this tension, stating: "To developers and technically minded people, it's this cool thing, but to illustrators it's very upsetting because it feels like you've eliminated the need to hire the illustrator."
In August 2022, Jason Allen entered an image titled "Theatre D'opera Spatial" in the digital arts category at the Colorado State Fair's fine art competition. The image, which depicted classical figures in an ornate hall looking through a circular viewport, won first place and a $300 prize. Allen had used Midjourney to generate the image, employing at least 624 text prompt iterations before editing the result in Adobe Photoshop and upscaling it with Gigapixel AI.
The award drew widespread controversy when it became public that the winning entry was AI-generated. Many traditional artists accused Allen of cheating and argued that AI-generated art should not compete alongside human-created work. Allen maintained that he was using Midjourney as a creative tool, similar to how other artists use digital software.
The controversy deepened when Allen applied for copyright protection for the image. On September 5, 2023, the U.S. Copyright Office Review Board issued a final determination that "Theatre D'opera Spatial" could not receive copyright registration. The board concluded that Allen's contribution (typing text prompts) was insufficient to establish human authorship, since the AI-generated elements dominated the final work. In 2024, Allen filed an appeal in U.S. District Court in Colorado, challenging the Copyright Office's decision.
In March 2023, a Midjourney-generated image of Pope Francis wearing a large white Balenciaga-style puffer jacket went viral across social media platforms, fooling millions of viewers who believed it was a real photograph. The image was created by a user named Pablo Xavier and posted to a Facebook group called AI Art Universe before spreading to Reddit and other platforms.
The incident was widely described as the first major case of mass-scale AI-generated misinformation. While the image contained telltale signs of AI generation (such as distorted hands and overly sharp skin textures), the overall quality was convincing enough to deceive casual viewers. Shortly after the incident, Midjourney discontinued its free trial program to curb misuse. The company also implemented content restrictions around certain prompts; for example, the word "pope" was temporarily restricted following the viral images.
Midjourney's use of copyrighted images as training data has become the subject of multiple high-profile lawsuits:
Midjourney has argued in its defense that its use of training data falls within fair use parameters. The company has also pointed out that the plaintiff companies themselves use generative AI tools, quoting Disney CEO Bob Iger's statement that "technology is an invaluable tool for artists."
These lawsuits are part of a broader wave of over 70 copyright infringement cases filed against AI companies as of 2025, and their outcomes may set significant legal precedent for the entire generative AI industry.
Beyond the Pope Francis incident, Midjourney's improving photorealism has raised ongoing concerns about its potential for creating deepfake images and visual misinformation. A mid-2023 algorithm update further enhanced the tool's ability to produce convincing depictions of real public figures. Midjourney maintains content moderation policies and employs approximately 40 moderators to monitor generated content. The platform prohibits the creation of adult content, gore, and generally disturbing or shocking imagery. Some prompts are blocked automatically by the system's content filters.
Midjourney V7 launched in alpha on April 3, 2025 and became the default model on June 17, 2025. It represented a complete rebuild from scratch rather than an incremental update. Key innovations included Draft Mode, which produces prototype images 10 times faster at half the cost; default model personalization that adapts outputs to individual user preferences; and a conversational editing mode on the web interface. V7 also brought significantly improved coherence for bodies, hands, and fine details.
Midjourney entered the video generation space on June 19, 2025, with its V1 Video model. The image-to-video workflow integrates with the existing image generation pipeline, allowing users to animate their still images into 5 to 21-second clips.
On August 24, 2025, Meta announced a partnership with Midjourney to license its image and video generation technology. The deal brings Midjourney's "aesthetic technology" into Meta's ecosystem, with plans to integrate AI-generated image and video tools into Facebook, Instagram, and WhatsApp. Midjourney confirmed it would continue operating independently despite the partnership, with Holz reaffirming the company's identity as "a community-backed research lab" with "no investors."
On March 17, 2026, Midjourney released the V8 Alpha preview on alpha.midjourney.com. The new model is the fastest to date, rendering standard jobs approximately 4 to 5 times faster than earlier versions. V8 introduces native 2K resolution images via the --hd parameter (eliminating the need for separate upscaling), a new --q 4 quality mode for additional coherence, and substantially improved prompt adherence, particularly for text rendering within images. Text placed in quotation marks within prompts is rendered with significantly improved accuracy, producing readable street signs, clean product labels, and legible typography.
The V8 Alpha also brings an upgraded web UI with a new grid view, settings sidebars, improved conversation mode, and moodboard support. As of its alpha release, V8 is available only on the alpha site and not yet accessible through Discord or the main website.