Changelog

Seedance 2.5 - the strongest engine in the studio writes the motion and the sound of a shot together, so clips arrive with their own audio. Up to 12 seconds, from 85 credits.

A cast of up to 10 - Seedance 2.5 Character takes up to ten reference photos at once (saved characters and uploads, cited automatically as @Image1-@Image10) and keeps every one of them consistent in the scene. Consent confirmation for photos of people.

Three new image engines - Nano Banana 2 (Google), Seedream 4 and Seedream 3 (ByteDance) join the image generator, each with an example image rendered from the same prompt so you can compare before spending a credit.

Image references - hand Seedream 4 or Nano Banana 2 a photo and its subject appears in the image you describe, with the same consent confirmation as video.

Pick from your library - wherever a reference photo can be attached, a popup now offers your Media Library images directly. No more downloading a file just to upload it again.

Choosing an engine no longer jumps you to the next tab - the example stays on screen and a pulsing dot points to the next step instead.

On phones, option tabs wrap onto a second row instead of scrolling sideways - nothing hides off the edge of the screen anymore.

The Flux 2 Pro tile now runs the real Flux 2 Pro model.

12 AI video models - seven new engines join the clip generator: LTX 2.3 Fast, PixVerse v6, Hailuo 2.3 and 2.3 Pro, Wan 2.7, Seedance 2.0 Mini, and Kling v3 Pro. From 25 credits per clip.

Character consistency - upload a photo of your character and the new reference model puts the same character into any scene you describe. 50 credits per clip, with a consent confirmation for photos of people.

Example clips on every model - the same prompt rendered by all engines, so you can compare quality before spending a credit.

Tier tabs in the model picker - Budget, Standard, Cinematic, and Ultra, so you choose a quality level first and a model second.

Kling v3 - the clip generator jumps two model generations and now produces native audio. 55 credits per clip.

Veo 3.1 Fast - AI clips now render in full 1080p at the same credit price.

Every registration is greeted the same way - welcome credits now always appear in your credit history, whichever way you signed up.

Talking Avatars - turn a character photo into a speaking presenter with lip-synced voiceover.

Background music at generation time - six styles, mixed and ducked under the voiceover.

Change your plan in-app - upgrades start immediately and charge only the pro-rated difference; downgrades take effect at the end of the paid period.

Captions land on the spoken words, scene pacing follows the voiceover, and ordered video length is enforced end to end.

Prompt-to-Video pipeline: script, scene images, voiceover and a rendered MP4 from a single prompt.

In-browser timeline editor with scene, voiceover and music tracks.

AI Image and AI Video generators.

Subscription plans and billing, with a monthly credit allowance.

Transactional emails — welcome, video ready, and password reset.

Complete CASTMINT visual redesign — dark, neon-gradient interface across the whole app.

Faster render queue and more reliable scene generation.

Media Library, AI Characters and Video Series drafts.

Early access: account creation, credits and the first end-to-end video generation.

30+ language support for scripts and voiceovers.