Midjourney launches its first video model

Background

Turning a still image into video means generating a sequence of frames that change over time while remaining visually coherent. Midjourney's V1 was the company's first model for this kind of task.^[1]

What happened

On 18 June 2025, Midjourney launched V1, its first video model.^[1] At launch it was web-only.^[1] The model animates images into 5-second clips, and those clips can be extended to about 21 seconds.^[1] A video job costs about eight times as much as an image job.^[1]

Key details

Each video job produced four 5-second videos.^[2] Each of those videos could be extended by about 4 seconds at a time, up to four times.^[2] Users could choose automatic animation, in which the system wrote the motion prompt, or manual animation with low and high motion settings.^[2]

Why it matters

Midjourney described real-time open-world simulation as its long-term goal.^[1] At launch, V1 was a web-only tool for short clips with video jobs priced at roughly eight times an image job.^[1] The company's stated long-term goal therefore sat alongside a much more limited released capability: animating images into short videos and letting users extend them step by step.^[1][2]