Midjourney V7 Launches With ‘Draft Mode’ That Costs Less
Midjourney has released a new version of its AI image generator, V7, which is its “smartest, most beautiful, most coherent model yet.”
Midjourney was once at the forefront of AI image generators — a controversial new industry that encroaches on the photography space — having been a very early product. However, it seemingly doesn’t have the influence it once did.
Perhaps that is because Midjourney is, as Tech Crunch describes, an “unusual company” since it has never raised any outside money. Or perhaps it is because of the lawsuits it is facing.
We're now beginning the alpha-test phase of our new V7 image Model. It's our smartest, most beautiful, most coherent model yet. Give it a shot and expect updates every week or two for the next two months. pic.twitter.com/Ogqt0fgiY7
— Midjourney (@midjourney) April 4, 2025
Midjourney’s V7 model is currently in its alpha phase. The company is touting a new feature called “Draft Mode” which costs less to use and is reportedly 10 times the speed. “It might be the best way to iterate on ideas ever,” Midjourney says.
Draft Mode works in “conversational mode” and can be controlled by the user’s voice, enabling them to think out loud “and let the image flow beneath you like liquid dreams.” Draft mode does produce lower-resolution images than standard mode.
One of the most exciting new features for our new V7 model is something we call "Draft Mode". Draft mode is half the cost and 10 times the speed and it might be the best way to iterate on ideas ever. Try it with voice, think out loud and let our ideas flow like liquid dreams. pic.twitter.com/ANfTMC6Ej1
— Midjourney (@midjourney) April 4, 2025
Midjourney V7 is the first model to have “personalization” switch on by default. To use this mode, the user must first rate 200 images. Midjourney uses this data to build a visual profile for the user.
Midjourney is best known as an app available on Discord but nowadays it also has a standalone website.
“V7 is much smarter with text prompts,” Holz says in an announcement on Discord. “Image prompts look fantastic, image quality is noticeably higher with beautiful textures, and bodies, hands, and objects of all kinds have significantly better coherence on all details.”
There are still features rolling out to V7, such as image upscaling and retexturing. These will arrive in the coming months.
“This is an entirely new model with unique strengths and probably a few weaknesses,” Holz writes on Discord. “We want to learn from you what it’s good and bad at, but definitely keep in mind it may require different styles of prompting. So play around a bit.”
