Release Notes - April 2026

Julia Enthoven -

New Models and Model Selection

As of this month, Kapwing AI now uses WAN 2.2 as the default video generator for free users, making AI video generation more accessible to people trying Kapwing without a paid plan. We also added GPT Image 1.5 for higher-quality outputs and Meta's Segment Anything for blur video. For music generation, Kapwing added support for including Minimax Music 2.5, Google Lyria Pro, and ElevenLabs music models.

Try out the WAN model and others using the new Model Selection model. We also cleaned up the Model Preferences modal by removing the Script to Video section, leaving Images, Video Clips, and Audio. In the Kapwing AI input box, click the model configuration button to set your preference for how Kai chooses which model to use for your generations. The credits estimate shows up before you confirm a video project generation:

  • Auto ($$): Chooses the best model based on your prompt, opting for more affordable models when reasonable.
  • Highest Quality ($$$-$$$$): Veo 3
  • Most affordable ($): WAN 2.2

Better Blur Video

Kapwing AI can now blur moving objects in videos from a text prompt, starting with common prompts like “blur faces” or “blur license plates.” The new SAM3-powered masking tool creates a linked mask layer. After generating in Kapwing AI, expand, feather, or remove the blur preview in the Studio.

Improved Kapwing AI Interface

In April, we added dozens of usability improvements to Kapwing AI, our prompt-to-edit surface. Results from the Clip Generator now includes a subtitles dropdown in preview, making it easier to inspect and adjust subtitles before moving forward with clips. Creators can also choose the aspect ratio of their output. To simplify new chat creation, they keyboard shortcut CMD+shift+O now opens a new Kapwing AI chat. Kapwing AI introduced lightboxes for audio and PDF uploads so that users can review their inputs..

Kai’s multi-project output experience was redesigned. Now, the transcript appears next to the result, and each project can be reviewed as it's own full width block.

Kapwing AI on mobile got several targeted fixes: screen does not zoom when copying a share link; the interface for edit-plan and confirm buttons responds better to a narrow device; and the notification banner layout takes up less space. We also tightened the mobile zero-state chat input and footer layout for narrow screens.

Export as AAC

The Kapwing Studio now supports AAC audio, music, or sound effect export. Audio-only exports can be downloaded in the .aac file format, which is the default encoding method for Mac computers and software. Find the '.ACC' file option in Export Project > Audio file format selector, under '.MP3'.

Other bug fixes and feature improvements

  • Removed legacy Seedance 1.0
  • Continued the migration of Team Assets, enabling users to filter by "Uploaded by Me" vs "Uploaded by the Team" in the Media Library area.
  • As of this month, Kapwing AI can modify just one generation in a project rather than re-generating every layer. This improved intelligence about the existing projects in your chat makes it easier to iterate in the chat interface.
  • New chats auto-scroll to the top of the Kapwing AI page, making room for the response.