Midjourney launches V7 with personalization and voice mode

Midjourney introduced the alpha version of the V7 model and opened testing access to all users. It features an entirely new architecture with improved handling of text and visual prompts, enhanced image clarity, and more accurate rendering of bodies, hands, and small objects.

The main innovation is default personalization. To activate it, users need to complete a simple setup that takes about five minutes. Personalization allows the system to better understand a user's tastes and visual preferences.

Another key feature is “Draft Mode.” In this mode, images are generated 10 times faster and cost half as much. The interface becomes “conversational,” enabling voice interaction with the model: users can replace a cat with an owl or change the time of day simply by saying it aloud. Draft image quality is lower, but they maintain stylistic consistency and are ideal for generating variants. Any image can later be “enhanced” or “varied” with full detail if needed.

The model is available in Turbo and Relax modes. Currently, only Turbo mode is operational, and it requires twice the resource cost compared to V6. Quality and speed improvements for the standard mode are expected later.

Some features like upscaling, editing, and retexturing are not yet available in the new model. Developers promise to update them soon. Weekly updates are also planned over the next two months. The most anticipated addition is a new tool for working with characters and objects based on V7.