Midjourney's Image Generator Gets First Major Update in 18 Months

After a year and a half of silence, the Midjourney team has unexpectedly announced version 7 of its image generator. V7 is currently in alpha testing via Discord but is already turning heads with its fresh approach to human-AI interaction.

One of the standout features is personalization. Before getting started, users are asked to rate 200 random images in just five minutes. This quick survey helps the algorithm learn your preferences — do you lean toward bold colors or muted tones? Realism or abstraction? Clean minimalism or intricate detail? The idea is to reduce the time spent fine-tuning prompts.

A new Draft mode makes editing more flexible. Instead of rewriting the whole prompt, you just add a clarification, and the AI adjusts the image while preserving the original concept. It can also remove unwanted elements like backgrounds or weather effects. An especially cool addition is voice input in Draft mode — you can simply speak your idea, and the AI instantly turns it into a sketch. Perfect for quick brainstorming sessions.

You can now also control how fast images are generated with two new modes:

Under the hood, there are noticeable improvements in detail: better textures, lighting, and text rendering, along with fewer awkward faces and hands. Some features from V6 — like advanced upscaling — haven’t made it over yet, but they’re in the pipeline.

For now, V7 is still in testing, and some tools may be a bit buggy. The team plans to release updates every two weeks, gradually improving stability. The big question remains: will personalization truly streamline the creative process? One thing is clear — Midjourney is betting on smart adaptation over raw power.