Zopia × SVAiFF
Create with AI. Be seen by the world.
Submission deadline ·
200 Zopia Universal Credits + 800 Vidu Credits
Credit eligibility and redemption follow the campaign terms.
View event
Zopia × YFCon
Zopia at YFCon 2026: connecting with creators and industry partners through AI creation.
Zopia exhibitor · Ended · 2026.10.08–10.09
Anaheim Convention Center · Hall A
Booth B14
Zopia × AIGC ASIA
Celebrate Eastern culture with AI and share new cultural visions with the world.
Gold awards in two tracks, ¥20,000 each
Register and receive 1,000 credits
Sep 1–Oct 11
More creative possibilities, more rewards
Create without limits · Earn without limits
Create with AI. Share your boundless imagination.
Great ideas deserve to be seen
CREATORS
MAKE A BRIGHTER WORLD
Invite friends to top up and earn cashback rewards.
Create together, grow together
TOGETHER
WE CREATE MORE
Invite friends or apply as an affiliate and earn rewards.
Connect with creators. Grow together.
CREATE TOGETHER
GROW TOGETHER
How HappyHorse 1.0 stacks up on benchmark ranking, audio-visual sync, and speed.
| Feature | HappyHorse 1.0 | Sora 2 | Veo 3.1 |
|---|---|---|---|
| Artificial Analysis Video Arena ranking (T2V & I2V) | #1 Elo | Top 5 | Top 5 |
| Native joint video + audio (single forward pass) | Yes — built in | Limited | Yes |
| Lip-sync languages | 7 (EN/ZH/Cantonese/JA/KO/DE/FR) | EN focus | Limited |
| 1080p generation time (single H100) | ~38 seconds | ~2–3 min | ~1–2 min |
| Reference-to-video (R2V) | Yes — dedicated endpoint | Limited | Yes |
| Free trial | Yes — starter credits | Limited | Paid |
HappyHorse 1.0 is Alibaba's flagship AI video model, launched on fal in April 2026 and ranked #1 on the Artificial Analysis Video Arena leaderboard for both Text-to-Video and Image-to-Video. Unlike pipelines that bolt audio onto video after the fact, HappyHorse uses a unified 40-layer self-attention Transformer that generates video and audio jointly in a single forward pass — no separate audio post-processing, no cross-attention modules. The result is natively synchronized output: lip-sync, footsteps, ambient sound, and dialogue all emerge from one prompt. Built for product promos, social content, and multi-shot sequences with consistent character identity.
Six capabilities that put HappyHorse at #1 on the public video arena.
A single 40-layer Transformer produces frames and audio together. Lip movement, footsteps, and ambient sound emerge synchronized — no post-dub, no manual alignment.
Native lip-sync for English, Mandarin, Cantonese, Japanese, Korean, German, and French. Write dialogue in any of them and HappyHorse mouths it correctly.
Text-to-video, image-to-video, reference-to-video, and video edit — all from the same model, exposed as four dedicated endpoints on Zopia's pipeline.
Upload up to 9 reference images of a character or product; HappyHorse holds visual identity across multi-shot sequences without retraining.
~38 seconds for a 1080p clip on a single H100. Iterate at the speed of conversation — critical for ad creative testing.
Top of the Artificial Analysis Video Arena leaderboard for both Text-to-Video and Image-to-Video as of April 2026 — the most rigorous public head-to-head video arena.
From a blank canvas to a synced clip in three steps.
Type a prompt, upload an image (I2V), or upload up to 9 references for reference-to-video. Mix and match — HappyHorse picks the right endpoint automatically.
Describe what you want to hear (dialogue in any of 7 languages, footsteps, ambient) and what you want to see (camera movement, action, lighting). Sound and visuals generate together.
Pick aspect ratio (16:9 / 9:16 / 1:1 / 4:3 / 3:4), duration (3–15s), and resolution (720p / 1080p). 1080p generates in roughly 38 seconds — run variants side by side.
HappyHorse takes audio cues seriously — be explicit about what should be heard. Best structure: subject + dialogue + ambient sound + camera + scene + style. Example: "A barista in a Tokyo cafe + says 'おはようございます' to camera + soft espresso machine hiss in background + slow push-in + warm morning light + cinematic." For lip-sync, write the literal line you want spoken in quotes — HappyHorse handles 7 languages natively. For reference-to-video, upload the character/product image and describe the scene; HappyHorse holds identity automatically. Avoid mood-only prompts ('make it cinematic'); always anchor with subject + action + sound.
From a single prompt to a synced 1080p clip — start generating in seconds.
Generate for FreeEverything you need to ship a synced clip — at a glance.
Same one-prompt experience, different specialties.