Midjourney releases new AI image model amid ChatGPT craze
What's the story
Midjourney, a pioneer in AI image generation, has launched its latest model, V7. The rollout started early Friday morning, marking the company's first new release in nearly a year. This comes just days after OpenAI's new image generator for ChatGPT took the internet by storm with its ability to create Ghibli-style images, among other styles.
Model features
V7 model: A new approach to AI image generation
Though the V7 model isn't specifically optimized for Ghibli-style images, it does produce visually appealing works. To use this new model, users have to create a Midjourney "personalization" profile by rating some 200 images. This profile customizes the model according to individual visual preferences and is automatically activated in V7. Once done, users can access the toggleable V7 on Midjourney's website or its Discord chatbot if they are members of Midjourney's Discord server.
Twitter Post
'V7 is our most coherent model yet'
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
Model performance
Midjourney CEO highlights V7's advanced capabilities
Midjourney's CEO David Holz called V7 a "totally different architecture." He said the new model excels with text prompts, delivers images with rich textures, and significantly improves coherence, making all details more accurate and visually appealing. The V7 model is available in two variants: Turbo and Relax. The Turbo variant is costlier to run but fuels a new tool called Draft Mode which creates images at 10 times the speed and half the cost of standard mode.
Company outlook
Future plans and challenges
Holz admitted that some features available in previous models aren't yet available in V7, including image upscaling and retexturing. He promised users that these features will be available soon. Holz also noted on Discord that this is an entirely new model with its own strengths and probably a few weaknesses. He urged users to play around with it as it may need different styles of prompting.