Midjourney, one of the first IA -based image generators on the web, published its first new IA image model in almost a year.
Double V7The model began to take place in Alpha around midnight is Thursday, occurs a week after Openai made its debut on a new image generator in the Chatppt which quickly became viral for its ability to create Ghibli- style photos.
Midjourney’s is not a model optimized by Ghibli, however – at least not officially – but it can nevertheless generate aesthetically pleasant works, at least to the dilettan eye of this journalist.
We are now starting the alpha-test phase of our new V7 image model. It is our smartest, most beautiful and coherent model to date. Give him a shot and wait for updates each week or two for the next two months. pic.twitter.com/ogqt0fgiy7
– Midjourney (@midjourney) April 4, 2025
To use it, you will first need to assess around 200 images to build a Midjourney “personalization” profile, if you have not already done so. This profile listens to the model to your individual visual preferences; V7 is the first Midjourney model to be referred by default.
Once you have done this, you can activate or deactivate V7 on the Midjourney website and, if you are a member of the Midjourney Discord server, on her Discord chatbot. In the web application, you can quickly select the model from the drop -down menu next to the “version” label.
Midjourney CEO David Holz described V7 as a “totally different architecture” in a Publish.
“V7 is […] Much smarter with text prompts, “said Holz in an advertisement on Discord.”[I]Mage’s prompts are fantastic, image quality is significantly higher with beautiful textures, and bodies, hands and objects of all kinds have a significantly better consistency on all details. »»
V7 is available in two flavors, Turbo (Costlier To Run) and relax, and feeds a new tool called Draft mode which makes the images at 10 times the speed and half of the cost of the standard mode. The images caused are of lower quality than images in standard mode, but they can be improved and returned in one click.
A number of Midjourney’s standard features are not yet available for V7, according to Holz, including image scaling and retacturation. They will arrive in the near future, he said, perhaps within two months.
“This is an entirely new model with unique forces and probably some weaknesses,” Holz wrote on Discord. “”[W]I want to learn from you what it is good and bad, but keep in mind that it may require different styles of incentive. So play a little.
In my brief test, V7 reasonably adhered to my prompts, although I did not have time to Really Put the model through the ringtone.
Midjourney is an unusual operation. Launched in 2022 by Holz, who co -founded the peripheral company of PC Leap Motion, he did not raise a penny outside the money.
At the end of 2023, Midjourney was would have Expecting to bring about $ 200 million in income. Recently, the company based in San Francisco said that it was establishing a material team to work on certain projects that it had not detailed, and that it continues to train models previously announced for the generation of videos and 3D objects.
The company is faced several legal proceedings Who accuse him of abandoning the rights of millions of artists by forming AI tools on the scraped images of the web without the consent of the creators of the images.
