A new image generation model is introduced by Midjourney after more than a year
Midjourney has rolled out its newest AI image generation model, named V7, marking its first major update in quite some time. Now available in alpha to users, V7 represents a complete rebuild from the ground up and focuses on two major areas of advancement—image quality and the overall creative workflow.
In terms of visual enhancements, the V7 model brings significantly improved image coherence and consistency, especially in areas that have traditionally posed challenges for AI, such as rendering hands, fingers, various body parts, and a wide range of objects. One of the standout improvements lies in the textures and surface materials within generated images. Skin now features more realistic details like fine wrinkles, and materials such as ceramics display subtle visual nuances that elevate the realism. These refinements address some of the most obvious flaws that typically reveal an image as AI-generated. While Midjourney is not claiming that its images can completely fool the trained eye, it does assert that many of the usual inconsistencies and visual artifacts have been noticeably reduced.
On the functional side, one of the most anticipated additions is the introduction of “Draft Mode.” Midjourney has announced across its communication platforms—including its blog, Discord server, and X profile—that this new mode offers a substantial increase in rendering speed, delivering outputs ten times faster than previous methods while also cutting the generation cost in half. However, these draft images are not intended for final presentation, as they come at a lower resolution and visual fidelity. The goal here is to enable users to rapidly explore concepts and iterate quickly before committing resources to producing a polished final image.
The V7 model includes two operational settings: turbo and relax. Turbo mode generates high-quality images at a faster pace but consumes twice the credits, while relax mode operates more slowly but is more credit-efficient. Notably absent is the standard mode familiar to long-time users. According to Midjourney, this standard mode is still in development and will be introduced later once further optimization is completed.
