Changelog
Follow up on the latest improvements and updates.
RSS
Introducing MAGI-1, the latest addition to the Video Generator.
Key advantages include:
- Excellent motion and video quality
- Strong understanding of prompts
- Low cost
Available in Video Generator
improved
Video Generator
Video Generator update
We’ve introduced two major updates to the Video Generator:
1. Redesigned interface
The Video Generator interface has been fully rebuilt to provide a more intuitive and streamlined user experience.
2. New video generation models
Six advanced video models have been added, offering a range of options for aspect ratios, video length, resolution, and frame rate.
Available models:
- Google Veo 2
- Runway Gen-4 Turbo
- Runway Gen-3 Turbo
- Kling 2.0 Master
- Kling 1.6 Pro
- Kling 1.6 Standard
- Minimax 01
Read guides here.
Use Video Generator here.
new
improved
AI Generator
New Image References for Flux [dev] and improved Model Trainer
We've added two new Image References to Flux [dev]:
1. Hard edges
- Perfect for preserving objects and original image composition while changing the style.2. Depth
- Perfect for generating images with the same depth as the original image.Image references also work with Model Trainer Characters and Styles, meaning you can combine your reference images with your custom-trained models.
Additionally, we’ve improved the Model Trainer’s training process. It now gives better results with longer, more creative prompts – and delivers more photorealistic and diverse images from shorter ones.

Key features:
- The new inpaintingmodel offers significantly higher quality and precision. The tool now offers 5 ways to mask an area, including a magical auto mask.
- Background replacementallows you to swap any background easily - just describe the desired background.
- You can easily restyleany image in seconds - just describe the desired style.
- The new Image Editor is fully optimized for mobile, so you can edit on the go.
- You can also apply filtersandadjust image parameters, such as brightness, contrast, or saturation.
The new Image Editor is available here.
We've prepared dedicated guides.
The new Image Editor will eventually fully replace the old one. The old Image Editor is still available under this link.
new
AI Generator
FLUX1.1 [pro] Ultra
Introducing FLUX1.1 [pro] Ultra to all plans - including free.
Key advantages:
- supports images up to 4MP
- the default resolution is 2048x2048px

improved
new
fixed
Image to Video -> Video Generator
Image to Video has been renamed to Video Generator, as it now also supports text-to-video.
What's new
- We have added several new modes of video generation:
- Standard: A general mode for both text-to-video and image-to-video creation.
- Live: Designed specifically to animate sketches and drawings.
- Subject: Capable of producing videos of a character by referencing an image.
- Director: A text-to-video mode featuring camera movements and zoom effects.
- We have rolled out our new Audio Generation Model. It enables the creation of videos enriched with sound.
The Image Variations tool lets you quickly create up to 10 high-quality, unique versions of any image. It works perfectly with photos, abstract art, logos, and anything you can imagine.
new
improved
Model Trainer replaces DreamBooth
A brand-new tool - Model Trainer replaces DreamBooth.
Key features:
- Upload images of any aspect ratio for training.
- Powered by FLUX [dev] to deliver sharper, more detailed results.
- The strength parameter gives you more control when generating images with your trained model.
- Combine models: Use Style and Character models in a single generation.
- Unlimited storage: Keep an unlimited number of trained models saved in your account.
- You can use up to 100 images for a single training.
- Upload .jpg, PNG, or WebP images for training.
- Flexible pricing: 1450 credits for up to 20 images used for training and 75 credits for each additional image.
Learn how to use Model Trainer - watch the video guide below.
You can now add tags to your images. Tags help you group your creations by themes, styles, or even moods. Once tagged, your images become easier to find, filter, and manage.
To learn how to use this new feature, watch the video guide below.
improved
Image to Video update
The current model has been upgraded to MiniMax (Hailuo AI).
What’s New:
1. Unlimited Resolutions:
Now, you can use any format directly from the AI Generator, and the video size will automatically adjust to match it.2. Optimized Prompts:
An additional model runs in the background, optimizing prompts for even better results.Note:
With this enhanced performance, processing times are around 10x longer, and videos now cost 600 credits per creation.Load More
→