Midjourney v7 is the image generation model that Midjourney released on April 3, 2025, its first major model update in close to a year, and it became the default version on the platform that June. It sharpens text and image prompt adherence compared to v6, with noticeably better hands, bodies, and small objects, and it introduced Draft Mode, a faster and cheaper generation setting meant for rapid exploration before committing to a full-quality render. V7 also added Omni Reference, for keeping a consistent character or object across generations, and turned on personalization by default, asking new users to rate around 200 images so the model can adjust its output to their taste.
Midjourney runs the model through Discord and its own web app rather than a public API, and pricing is subscription-based. It competes with tools like Stable Diffusion, Flux, and OpenAI’s image models, and Midjourney has kept its edge mainly on aesthetic quality and stylization rather than photorealism or prompt-following precision, though v7 narrows that gap somewhat.