Short-form creators
Scenario: Generating B-roll and reaction clips for social posts
Outcome: Daily content cadence without a camera shoot
by Alibaba ATH Innovation Division
Alibaba's 2D video generation model from the same ATH Innovation team as Happy Oyster, focused on producing short-form clips from text and image prompts.

Happy Horse is a text-to-video / image-to-video model from Alibaba ATH Innovation Division. It is currently in public stage (since 2026-03-10).
Produces short video clips directly from natural-language prompts.
Animates a still image into a short clip with controllable camera motion.
Provides higher-level motion intent (camera pan, subject action) on top of the base prompt.
Stitches multiple generated shots into a single coherent sequence.
These claims are drawn from Alibaba ATH Innovation Division's own positioning and should be verified against hands-on testing once general access opens.
| Output type | 2D video clips ✓ |
|---|---|
| Max clip length | Not officially documented ? |
| Resolutions | 720p / 1080p tiers reported ~ |
| API | Public access via tryhappyhorse.xyz ✓ |
| Pricing | Tiered, details not finalized ~ |
Scenario: Generating B-roll and reaction clips for social posts
Outcome: Daily content cadence without a camera shoot
Scenario: Producing product showcase loops from a single hero image
Outcome: Ad creative variations in minutes per SKU
| vs | On | Happy Horse | Them |
|---|---|---|---|
| Happy Oyster | Output paradigm | 2D video clip | Interactive 3D world |
| Sora | Provider ecosystem | Alibaba ATH stack | OpenAI stack |
| Kling | Geographic positioning | Alibaba-aligned distribution | Kuaishou-aligned distribution |
Alibaba's 2D video generation model from the same ATH Innovation team as Happy Oyster, focused on producing short-form clips from text and image prompts.