This release adds the most talked-about AI video model of 2026 to the Genvid platform, alongside meaningful upgrades to the keyframe editor, expanded ComfyUI integration, and quality-of-life improvements throughout.
Alibaba Happy Horse: The Current Number One, Inside Your Pipeline
On April 7, 2026, Alibaba’s video model “HappyHorse” appeared on the Artificial Analysis Video Arena leaderboard and immediately claimed the number one spot for both text-to-video and image-to-video generation. Creating video and audio that feels designed together rather than synchronized after the fact, it supports generation of up to 15 seconds of 1080p video with multiple shots, and delivers synchronized audio-visual output including ambient soundscapes and emotionally expressive lip-synced dialogue in seven languages: English, Mandarin, Cantonese, Japanese, Korean, German, and French.
Both Text to Video and Image to Video modes are available in Genvid now. Generate from a text prompt or animate any image in your asset library directly into a shot.
Keyframe Editor Enhancements: Move, Copy, and Upload Without Leaving the Editor
The Genvid keyframe editor now supports moving, copying, and uploading keyframes directly, keeping your creative flow intact and your keyframe library organized without round-trips between different parts of the interface. Invalidated keyframes are now visible in the editor with signed URLs, so the full picture of your keyframe state is always in front of you rather than hidden until something breaks.
ComfyUI on the Keyframe Page
ComfyUI workflows can now be run directly from the keyframe editing page, with full C2PA provenance applied to every image and video output. Bring your custom generation pipelines into the keyframe workflow without switching contexts, and every result carries the same cryptographically signed provenance record as any other asset generated in Genvid.
Cleaner Model Selection and Asset Management
Provider names now appear directly in the model dropdown, making it immediately clear which company supplies each model without cross-referencing documentation. Asset types can now be changed after upload, removing a friction point that required deleting and re-uploading assets to correct a misclassification.