Midjourney

Midjourney is an AI image and video generator with a reputation for producing artwork that looks unlike anything else in the category. The aesthetic is distinctive enough that creative professionals can often identify a Midjourney output on sight: richly textured, cinematically composed, with a quality some describe as dreamlike photorealism. Whether that's a strength or a limitation depends entirely on what you're trying to make.

The company was founded in 2021 by David Holz, who previously co-founded Leap Motion (acquired in 2019) after research stints at NASA's Langley Research Center and the Max Planck Institute. Midjourney operates with roughly 40 employees, has never raised external venture capital, and reached profitability within its first year. Revenue has scaled to well over $200 million annually on that lean basis.

V7, released in April 2025, was a full architectural rebuild. Prompt understanding improved by around 35 percent compared to V6, meaning you need fewer modifiers to get useful output. Anatomical accuracy improved by roughly 40 percent, though hands and faces still fail occasionally. Turbo Mode generates images at approximately ten times the standard speed at half the compute cost. Video generation arrived in June 2025, producing five to twenty-one second clips from static images with temporal consistency that keeps character identity and scene physics stable across the sequence. A V8 Alpha launched in March 2026 and is available to subscribers on alpha.midjourney.com.

The web interface has matured considerably. A Canvas mode lets you drag, drop, and outpaint images spatially rather than relying on Discord bot interactions. Discord is still available, but the web experience is now the primary one for most users.

For the use cases Midjourney is built for, keeper rates reflect genuine quality. Fantasy and concept art lands at around 90 percent usable output. Social media graphics and mood boards sit around 85 percent. Architecture and interior visualisation around 75 percent. Text rendering is where it fails consistently: legible text appears in roughly 10 percent of outputs, and product photography is unreliable at around 40 percent keeper rate. Character consistency across multiple scenes sits at 50 percent even with V7's improvements.

The honest caveats on the business side are significant. Every image generated is public by default. Stealth Mode, which keeps your output private, is only available at the Pro tier and above, meaning the effective starting price for commercial confidentiality is considerably higher than the entry plan suggests. There is no free tier, and the trial Discord credits that existed in earlier years are gone. To test the tool at all, you pay.

Customer support runs through Discord and documentation, with no ticketing system, no live chat, and no phone access. Trustpilot reviews consistently flag billing disputes and unanswered queries as the primary frustrations. In June 2025, Universal Pictures and Disney filed a copyright infringement lawsuit against Midjourney alleging unauthorized use of copyrighted training data, a case still ongoing as of early 2026.

For designers, concept artists, and marketers who prioritize aesthetic quality over pixel-perfect accuracy, Midjourney produces output that competitors at a similar price point consistently struggle to match in creative direction. For anyone who needs reliable text in images, consistent characters across a project, or genuinely usable product photography, the tool's weaknesses will surface fast.

Midjourney: Pros & Cons

| Pros (The Wins) | Cons (The Friction) |

| :--- | :--- |

| Artistic Quality: 90% keeper rate for fantasy and concept art output. | No Free Tier: No trial. $10 minimum to test the product at all. |

| V7 Architecture: 40% fewer anatomical errors; 35% better prompt accuracy. | Public By Default: All images public unless on Pro plan ($60+/month). |

| Video Generation: 5-21 second clips with stable character consistency. | Text Rendering: Only ~10% of text attempts produce legible output. |

| Self-Funded & Profitable: No VC pressure; focused on product, not fundraising. | Customer Support: Discord-only; billing disputes go largely unanswered. |

| Web Canvas: Spatial outpainting interface; no longer Discord-dependent. | Copyright Lawsuit: Disney/Universal suit filed June 2025, still ongoing. |

Features

Text-to-image generation (V7 default, V8 Alpha available)
Text-to-video: 5-21 second clips with temporal consistency
Web Canvas for spatial outpainting and editing
Turbo Mode: approx. 10x faster generation at half compute cost
Personalization system that learns user style preferences
Omni Reference for style and character consistency
Depth and Mesh parameters for 3D workflow integration
Draft Mode for rapid iteration
Stealth Mode on Pro and Mega plans
Discord bot interface alongside web
Four paid tiers: Basic, Standard, Pro, Mega