Midjourney V7: A New GPT-4o image generation competitor?
Created on April 4|Last edited on April 4
Comment
Midjourney has launched V7, its first new image generation model in nearly a year, marking a notable update for one of the most widely used AI art tools on the web. The rollout began around midnight Eastern time on April 3, 2025, and follows closely on the heels of OpenAI’s viral Ghibli-style image generator update to ChatGPT. While V7 isn’t officially designed to mimic Ghibli aesthetics, it produces visually appealing artwork that seems more refined than previous versions.
Personalization and Model Access
A key change with V7 is the automatic activation of personalization. Users must rate about 200 images to complete a visual profile that tunes the model to their aesthetic preferences. Once that's done, users can switch V7 on or off via Midjourney’s website or through its Discord bot. In the web app, the model is selectable via a simple dropdown menu. This is the first time personalization has been on by default, hinting at a deeper shift toward adaptive AI tools that tailor themselves to individual users.
Technical Improvements and Features
David Holz, Midjourney’s CEO, called V7 a “totally different architecture,” emphasizing improved handling of text prompts and better visual coherence. Compared to earlier versions, it generates more accurate details for bodies, hands, objects, and textures. The model comes in two operating modes: Turbo, which runs faster but costs more, and Relax, which is slower and cheaper. One standout addition is Draft Mode, which renders previews at ten times the speed and half the cost of standard images. These drafts can then be enhanced with a click, streamlining the creative workflow.
Some capabilities are still missing from V7, including image upscaling and retexturing, though Holz said those features will be added within two months. He also noted that V7 may respond better to new prompting techniques, suggesting users experiment to get the best results.
Company Strategy and Legal Pressures
Midjourney continues to run without any outside funding, an unusual stance in the AI space. Founded in 2022 by David Holz, who previously co-founded Leap Motion, the company reportedly neared $200 million in revenue by late 2023. It has recently begun building a hardware team for unspecified projects and is also developing tools for video and 3D object generation.
However, like many other generative AI firms, Midjourney faces legal scrutiny. The company is a target in lawsuits that claim its models were trained using copyrighted art scraped from the internet without permission. The outcome of these cases could have a significant impact on the future of AI art platforms and their data sourcing practices.
Early Impressions and Looking Ahead
Initial impressions of V7 suggest a more responsive and visually capable model, though it’s still early. Features are rolling out gradually, and Midjourney is soliciting feedback from users to refine the system. As generative art continues to evolve, V7 positions Midjourney to stay competitive, even as legal and ethical challenges mount.
Source:
Add a comment
Tags: ML News
Iterate on AI agents and models faster. Try Weights & Biases today.