TL;DR
HappyHorse 1.0 from Alibaba — the #1 ranked AI video model on the Artificial Analysis Video Arena (1332 Elo T2V, 1391 Elo I2V) — is now live on Renderful. Four variants: T2V, I2V, R2V (up to 9 reference images), and Video Edit (natural-language editing). Native 720P at $0.14/sec or 1080P at $0.28/sec, with 3–15 second durations. If you're already on Renderful, it's a one-line model ID change.
We're excited to announce that HappyHorse 1.0 by Alibaba is now available through the Renderful API. It's the highest-ranked AI video model on the independent Artificial Analysis Video Arena for both text-to-video and image-to-video — and on Renderful, you get all four HappyHorse variants behind a single REST endpoint with no Alibaba account required.
What HappyHorse 1.0 Brings to the API
HappyHorse 1.0 is Alibaba's flagship video model. Here's what it adds to Renderful:
#1 Ranked on Artificial Analysis Video Arena
HappyHorse 1.0 holds the top spot for both text-to-video (1332 Elo) and image-to-video (1391 Elo) on the independent Artificial Analysis benchmark — beating Sora 2, Veo 3, and Kling for realistic motion and visual coherence in head-to-head human preference voting.
Four Variants, One API
T2V (text-to-video), I2V (image-to-video with strict image consistency), R2V (reference-to-video with up to 9 reference images), and Video Edit (natural-language editing of existing footage while preserving motion).
Native 1080P Output
720P and 1080P generated directly by the model — no upscaling step, real spatial information per pixel. Choose 16:9, 9:16, 1:1, 4:3, or 3:4 aspect ratios.
Realistic Motion Dynamics
Hair flows, cloth drapes, water splashes with weight, and subjects walk with proper biomechanics. The motion realism is what wins the Artificial Analysis matchups.
Flexible Duration
3 to 15 second clips — short enough for ads and social, long enough for cinematic sequences. Pricing scales linearly with duration.
API Integration
HappyHorse 1.0 follows Renderful's standard generation format. If you're already integrated, it's a one-line model ID change:
{
"model": "alibaba/happyhorse-1.0-t2v",
"input": {
"prompt": "A hyper-realistic dew-covered orange
on a marble pedestal — a fluid-morph effect
dissolves the skin into swirling liquid that
spirals upward, solidifying into a sleek glass
bottle of orange juice. Cinematic slow-motion,
macro 8K, vibrant splash physics.",
"aspect_ratio": "16:9",
"duration": 5,
"resolution": "1080P"
},
"webhook": "https://your-app.com/webhook"
}Image-to-Video (I2V)
Animate any reference image as the first frame with strict consistency to the source:
{
"model": "alibaba/happyhorse-1.0-i2v",
"input": {
"image": "https://example.com/first-frame.jpg",
"prompt": "Camera dollies forward as the subject
turns slowly toward the light, hair drifting
in a gentle breeze.",
"duration": 5,
"resolution": "1080P"
},
"webhook": "https://your-app.com/webhook"
}Reference-to-Video (R2V)
Combine up to 9 reference images in a single shot. Reference each in your prompt with @Image syntax:
{
"model": "alibaba/happyhorse-1.0-r2v",
"input": {
"prompt": "A cool wedding dance scene between
@Image 1 and @Image 2 in the venue from
@Image 3, golden-hour lighting.",
"reference_image_urls": [
"https://example.com/character-1.png",
"https://example.com/character-2.png",
"https://example.com/venue.png"
],
"duration": 8,
"resolution": "1080P"
},
"webhook": "https://your-app.com/webhook"
}Video Edit
Edit existing footage with natural language while preserving the original motion dynamics:
{
"model": "alibaba/happyhorse-1.0-video-edit",
"input": {
"video_url": "https://example.com/source.mp4",
"prompt": "Make the horse-headed character wear
the striped sweater from @Image 1.",
"reference_image_urls": [
"https://example.com/striped-sweater.png"
],
"resolution": "1080P"
},
"webhook": "https://your-app.com/webhook"
}Full API reference in the Renderful docs.
Pricing
All four variants share the same simple per-second pricing:
| Variant | 720P | 1080P | Best For |
|---|---|---|---|
| HappyHorse T2V | $0.14/sec | $0.28/sec | Cinematic prompts, ads, social |
| HappyHorse I2V | $0.14/sec | $0.28/sec | Animate stills, brand-faithful motion |
| HappyHorse R2V | $0.14/sec | $0.28/sec | Multi-character scenes, branded narrative |
| HappyHorse Video Edit | $0.14/sec | $0.28/sec | Natural-language edits, motion preservation |
A standard 5-second 720P generation runs $0.70. Full pricing details.
How It Compares
HappyHorse 1.0 alongside other video models on Renderful:
| Feature | HappyHorse 1.0 | Seedance 2.0 | Kling 2.6 |
|---|---|---|---|
| Artificial Analysis Rank | #1 (T2V & I2V) | Top 5 | Top 5 |
| Max Resolution | 1080P (native) | 720p | 1080p |
| Multi-Reference | Up to 9 images | 9 img + 3 vid + 3 audio | 1 |
| Natural-Language Video Edit | Yes | No | No |
| Max Duration | 15s | 15s | 10s |
| Base Cost (5s 720P) | $0.70 | $0.83+ | $0.50+ |
Get Started
- 1Get an API key — create a Renderful account if you don't have one. Free credits included.
- 2Try the playground — test prompts and references directly on the model page before writing code.
- 3Integrate — use any of the four model IDs:
alibaba/happyhorse-1.0-t2v,alibaba/happyhorse-1.0-i2v,alibaba/happyhorse-1.0-r2v, oralibaba/happyhorse-1.0-video-edit. Results delivered via webhook.
Start Using HappyHorse 1.0
The #1 ranked AI video model on Artificial Analysis — text-to-video, image-to-video, 9-reference R2V, and natural-language video editing through a single API call.