Generate 3 to 15 second HappyHorse videos from text or one image in 720p or 1080p. Use multi-shot prompts and audio output for product demos, short ads, ASMR scenes, and storyboards; choose HappyHorse 1.1 when you need more aspect ratios or a larger reference upload.
HappyHorse is an Alibaba-ATH video model family for text-to-video and image-to-video creation. When HappyHorse 1.0 entered the Artificial Analysis Video Arena in April 2026, it placed first for no-audio text-to-video and image-to-video, and second in the matching audio-enabled arenas.
On Toolaze, choose HappyHorse 1.0 or 1.1 to create 3 to 15 second, 720p or 1080p clips from text or one image. Both versions suit product motion tests, short ads, character beats, tactile ASMR scenes, and multi-shot storyboard drafts.
AI video model
| Performance Metric | Toolaze Specification |
|---|---|
| Developer | Alibaba-ATH HappyHorse video model family |
| April 2026 launch benchmark | Placed first for no-audio text-to-video and image-to-video, and second in the matching audio-enabled Artificial Analysis Video Arena tests |
| Toolaze inputs | Text-to-video or image-to-video with one first-frame image |
| Output | 3 to 15 seconds, 720p or 1080p, 24 fps MP4 |
| Audio | Audio output with prompts for dialogue, ambience, music direction, and tactile sound effects |
| Prompt language | Alibaba Cloud documentation accepts prompts written in any language; it does not publish a fixed lip-sync language count or accuracy guarantee |
| Multi-shot direction | Write a clear shot order and repeat the subject, product, wardrobe, or scene details that must remain consistent |
| Aspect ratios | HappyHorse 1.1 adds 4:5, 5:4, 9:21, and 21:9 to the core 16:9, 9:16, 1:1, 4:3, and 3:4 set |
| Starting Credits | HappyHorse 1.1 starts at 135 Credits on Toolaze; HappyHorse 1.0 starts at 168 Credits |
Start with HappyHorse 1.1 for the widest ratio set and a 20MB image limit. Choose the base HappyHorse model when a simpler 10MB reference workflow is enough.
Use text-to-video for a fresh concept, or upload one reference image when identity, product shape, costume, or scene layout needs to stay recognizable.
Select 720p or 1080p, choose a supported duration, then describe the camera movement, shot order, subject action, lighting, and native audio cues.
Review the clip for motion drift, reference fidelity, and sound timing. Re-run with a shorter shot list or stronger preservation notes when the result feels too broad.
Use these as practical starting points, then set duration, resolution, and aspect ratio in the Toolaze controls.
Product Reveal
Create a premium product video for a matte black portable speaker on a stone table. Start with a slow side push, move into a close detail of the fabric grille, then reveal a soft blue light pulsing under the logo. Keep the product shape stable, add gentle bass vibration, room tone, and crisp button clicks.
Reference Character Beat
Animate the uploaded character image into a short dramatic moment. The character turns toward a bright window, coat fabric shifts with the movement, hair reacts to a light breeze, and the camera moves from shoulder detail to a calm face close-up. Preserve facial features, clothing colors, pose language, and background mood.
ASMR Texture Scene
Create a macro ASMR clip of translucent jelly cubes being pressed with a small metal spoon on a ceramic plate. Show slow compression, tiny surface ripples, droplets catching the light, and the spoon lifting away. Add soft squish sounds, plate contact, tiny taps, and no music.
Multi-Shot Lifestyle Ad
Create a lifestyle ad for a lightweight running jacket at sunrise. Shot one shows fabric moving in the wind, shot two follows the runner from the side, and shot three holds on the zipper detail as sunlight hits the reflective strip. Keep the jacket color and logo placement consistent, add footsteps, wind, and city ambience.
HappyHorse responds best when the prompt separates the version choice, the preserved subject, the motion path, and the sound layer.
Weak prompt
Make a fashion clip.
Weak
Make a fashion clip.
Better
Create a three-shot fashion clip for a silver jacket. Start with a shoulder detail, cut to a walking side view, then end on a close-up of the sleeve texture. Keep the jacket shape and color consistent, use soft sunrise light, and add fabric rustle plus quiet footsteps.
Weak prompt
Animate this image.
Weak
Animate this image.
Better
Animate the uploaded portrait with a slow camera push and a subtle head turn. Preserve face shape, hairstyle, jacket color, background layout, and lighting direction while adding natural breathing and light wind.
Weak prompt
Add ASMR sounds.
Weak
Add ASMR sounds.
Better
Create a close macro scene of a soap bar being shaved with a thin blade. Add crisp slicing, tiny flakes landing on ceramic, soft hand movement, and no background music.
Turn packaging, devices, beauty products, and apparel ideas into short motion drafts with stable shapes and clear lighting.
Upload one image and animate the subject while preserving the person, object, outfit, or scene layout.
Describe slicing, pressing, pouring, tapping, and fabric-contact scenes with matching native audio.
Build a compact shot order for TikTok, Reels, Shorts, and campaign testing without a full editing timeline.
Create expressive moments with camera movement, mood, costume continuity, and controlled body motion.
Test scene order, light direction, motion rhythm, and sound ideas before commissioning final video work.
Choose the HappyHorse version by input size, ratio needs, and how much output control the scene requires.
| Decision point | HappyHorse 1.1 | HappyHorse |
|---|---|---|
| Creation modes | Text-to-video and image-to-video with one image reference. | Text-to-video and image-to-video with one image reference. |
| Reference limit | One uploaded image up to 20MB. | One uploaded image up to 10MB. |
| Aspect ratios | Includes 16:9, 9:16, 1:1, 4:3, 3:4, plus 4:5, 5:4, 9:21, and 21:9. | Supports the core 16:9, 9:16, 1:1, 4:3, and 3:4 set. |
| Duration and output | 3 to 15 seconds in 720p or 1080p. | 3 to 15 seconds in 720p or 1080p. |
| Audio and shot control | Native audio and multi-shot prompting for product, story, and ASMR direction. | Native audio and multi-shot prompting for simpler drafts. |
| Starting Credits | Starts at 135 Credits on Toolaze. | Starts at 168 Credits on Toolaze. |
| Choose it when | You need the broadest HappyHorse ratio support or a larger reference upload. | You want the base HappyHorse model and can stay within the smaller reference limit. |
Use HappyHorse when scene continuity and version-level ratio control matter more than maximum reference count.
| Model choice | Good fit | Check before generating |
|---|---|---|
| HappyHorse | Alibaba text-to-video or image-to-video with one reference, native audio, and version-level ratio choices. | Confirm whether HappyHorse 1.1 or the base HappyHorse model fits the file size and aspect ratio you need. |
| Wan 2.6 | Sound-led Wan clips, native audio, and familiar 5/10/15 second timing. | Use it when you do not need the extra HappyHorse 1.1 wide or tall ratio set. |
| Kling 3.0 | Polished cinematic motion, product clips, and creator shorts with broader Kling model familiarity. | Check whether Kling motion style fits the brand before spending higher-quality runs. |
| Grok 1.5 Video | Fast native-audio drafts and broad creative exploration. | Use it for fast ideation when exact HappyHorse version controls are less important. |
HappyHorse 1.1 supports a larger 20MB image reference limit and adds wider ratio coverage including 4:5, 5:4, 9:21, and 21:9. The base HappyHorse model keeps the core ratios and a 10MB image limit, which is enough for simpler reference-led drafts.
When it entered the Arena in April 2026, HappyHorse 1.0 placed first for no-audio text-to-video and image-to-video, and second in the matching audio-enabled tests. This is a dated launch benchmark, not a permanent current position.
On Toolaze, HappyHorse creates 3 to 15 second MP4 clips in 720p or 1080p. Alibaba Cloud documents the HappyHorse family output at 24 fps.
Alibaba Cloud documents HappyHorse prompts as accepting any language. It does not publish a fixed lip-sync language list or an accuracy guarantee, so the safest approach is to test the exact spoken language and dialogue style before a final render.
Yes. Write the prompt as a short shot sequence with one action per beat, and include what should remain consistent across cuts, such as product shape, outfit color, logo placement, or character identity.
Yes. Add concrete audio cues such as fabric rustle, footstep rhythm, camera-room ambience, blade contact, or soft product sounds. Avoid generic wording like good audio when the sound needs to match the motion.
It can work well for ASMR-style material clips when the prompt names the texture, close camera distance, contact sounds, and no-music preference. It is best for short tactile scenes rather than long continuous ASMR sessions.
Yes. Use image-to-video when you need the person, product, costume, or layout to remain recognizable. HappyHorse accepts one reference image, with the file size limit depending on the selected version.
HappyHorse 1.1 starts at 135 Credits and the base HappyHorse 1.0 model starts at 168 Credits. The final amount depends on the selected duration and resolution and is shown before generation.
Use Wan 2.6 for text or one image, 5/10/15 second timing, native audio, and balanced Wan-style motion.
Create cinematic Kling video drafts when you want a familiar motion model and polished social clips.
Explore fast native-audio concept videos for broad ideation and social-ready drafts.