A new video model dropped at #1 on the leaderboard 👀 It's called HappyHorse-1.0, and it's currently leading in both text-to-video and image-to-video. From my testing, it's particularly good at multi-shot videos and following detailed directions 👇
视频模型
HappyHorse 1.0（阿里巴巴出品）- Artificial Analysis 排行榜第一的AI视频模型，单流Transformer统一生成
阿里巴巴 HappyHorse 1.0 在 Artificial Analysis AI 视频榜单（2026年4月）位居榜首。约150亿参数的单流 Transformer 架构，统一支持文生视频、图生视频、参考生视频、视频编辑与视频延展。可生成 720p/1080p、3-15 秒、多种宽高比的视频。
公开
Live PipelineTake 01 / 02
文生视频展示
一句话生成电影级画面。HappyHorse 将文字提示渲染为流畅写实的场景，光影丰富、运镜灵动，呈现院线大片般的质感。
Source Feeds00 Inputs
Pipeline
00%
Program · On AirAI · Generated
Output
Transcript · 01
怀旧卡带
Live PipelineTake 01 / 01
图生视频展示
看 HappyHorse 如何为静态图像注入电影级运镜、忠实的角色保持和自然的光影过渡，呈现多种场景。
Source Feeds01 Inputs
Pipeline
00%
Program · On AirAI · Generated
Output
Transcript · 01
参考图转视频
Live PipelineTake 01 / 01
参考图生成视频展示
看 HappyHorse 如何将多张参考图融合为一段连贯的视频。每段画面都基于 2-3 张源图,在运动生成前锁定角色形象、服饰与场景风格。
Source Feeds01 Inputs
Pipeline
00%
Program · On AirAI · Generated
Output
Transcript · 01
沙漠探险家
Live PipelineTake 01 / 01
视频编辑案例展示
基于提示词的视频编辑，支持多图参考。可替换物体、重塑场景、调整风格，同时完整保留原始动态。
Source Feeds01 Inputs
Pipeline
00%
Program · On AirAI · Generated
Output
Transcript · 01
场景换成时装T台
Live PipelineTake 01 / 01
视频续写展示
使用 HappyHorse 无缝延长已有视频，模型可保留原有的运动、光线与主体特征，让新画面如同一镜到底般自然衔接。
Source Feeds01 Inputs
Pipeline
00%
Program · On AirAI · Generated
Output
Transcript · 01
雨巷续写
HappyHorse 1.0 在 X 上的热门评价
看看创作者们如何评价新晋 #1 AI 视频模型 HappyHorse 1.0。
We’ve added a new pseudonymous video model to our Text to Video and Image to Video Arenas.‘HappyHorse-1.0’ is currently landing in the #1 spot for Text and Image to Video (No Audio) and the #2 spot for Text and Image to Video (With Audio). Further details coming soon. Example Show more
HappyHorse is still in the stable. 🐴 No official website yet — anything you've seen out there isn't us. We're part of Alibaba's ATH AI Innovation Unit, and when we're ready, you'll know. Stay tuned. #HappyHorse #Alibaba #ATH
Congratulations to @HappyHorseATH for securing #1 and #2 across the Artificial Analysis Video Arena leaderboards! A significant milestone, and the journey continues.🚀 #AlibabaAI #HappyHorse
Revealing HappyHorse-1.0 as the latest video model from Alibaba! @HappyHorseATH has landed in #1 or #2 across all of the leaderboards in the Artificial Analysis Video Arena. In our ‘without audio’ leaderboards, HappyHorse-1.0 is comfortably in first place. In our ‘with audio’
Revealing HappyHorse-1.0 as the latest video model from Alibaba! @HappyHorseATH has landed in #1 or #2 across all of the leaderboards in the Artificial Analysis Video Arena. In our ‘without audio’ leaderboards, HappyHorse-1.0 is comfortably in first place. In our ‘with audio’ Show more
New analysis: Alibaba’s HappyHorse topped the AI video leaderboard last week. The timing matters more than the score. It appeared during peak customer frustration with Bytedance’s pricing. Alibaba Cloud sales teams were already approaching Volcano Engine’s enterprise clients.
An anonymous model called HappyHorse appeared on Artificial Analysis’s Video Arena on April 7. Three days later, Alibaba claimed it. The model scored above Seedance 2.0 in both image-to-video and text-to-video categories. But the commercial logic matters more than the benchmark.
WHO IS HAPPYHORSE? IS IT WAN 2.7 VIDEO? Over the weekend a new MYSTERY AI VIDEO MODEL appeared on Artificial Analysis AI Video Arena as HAPPY HORSE V1 with a V2 variant. It definitely appears to be from Asia. If it is WAN 2.7 it is sizable leap from 2.6 Is it a Seedance 2.0 Show more
HappyHorse-1.0 is now live on Arena! 🚀 Early evals show exceptional performance in Video Edit. We are now in the final optimization sprint for the official launch in 2 weeks. We invite the community to get early access and test our capabilities at arena.ai. 🐎✨
🚀Exciting progress ahead. Can’t wait to see what @HappyHorseATH unlocks for creators in video edit! #AlibabaAI #HappyHorse
HappyHorse-1.0 is now live on Arena! 🚀 Early evals show exceptional performance in Video Edit. We are now in the final optimization sprint for the official launch in 2 weeks. We invite the community to get early access and test our capabilities at arena.ai. 🐎✨
Thrilled to see HappyHorse‑1.0 land #1 in the Video Edit Arena! 🎉@HappyHorseATH #AlibabaAI #HappyHorse
New video model HappyHorse-1.0 by Alibaba-ATH debuts at #1 in Video Edit Arena. It scores 1299, leading Grok Image Video by +42 points and Kling o3 Pro by +48 points. Video editing is an emerging frontier capability for video models, and only a small number of models support
YouTube 上的 HappyHorse 1.0
观看展示 HappyHorse 1.0 视频生成能力的演示和教程。
- Happy Horse 1.0 Takes the Lead! | API Tutorial - fal
- Happy Horse 1.0 Is Crushing SeeDance 2.0 (New #1 AI Model) - AI Factory
- Happy Horse 1.0 Just Destroyed Seedance 2.0 - Ai Chemy
- The Honest Truth About Happy Horse 1.0 (Watch Before You Buy) - AI Filmmaking Academy
- Happy Horse 1.0 is DESTROYING Seedance 2.0… This is Crazy 🤯 - AI Biceps
- Happy Horse: The Seedance Killer? Plus BIG 4k AI Video News! - Theoretically Media
YouTube 上的 HappyHorse 1.0
观看展示 HappyHorse 1.0 视频生成能力的演示和教程。
Reel · Specifications
什么是 HappyHorse 1.0
阿里巴巴在全球排行榜上排名第一的 AI 视频模型。
- · 01第1第1排行榜排名
- · 02150亿参数
- · 031080p分辨率
- · 043-15秒时长
HappyHorse 1.0 是一款约 150 亿参数的单流 Transformer 模型，于 2026 年 4 月登顶 Artificial Analysis 视频排行榜。
Reel · Capabilities
HappyHorse 1.0 功能特性
探索 HappyHorse 1.0 单流变换器视频生成模型登顶排行榜的强大能力
- Feature 01 / 08
排行榜第一
在 Artificial Analysis 视频生成排行榜上位列第一，在运动保真度、提示词遵循度和整体视觉质量等基准测试中超越知名竞品。
- Feature 02 / 08
单流变换器架构
采用高效的单流变换器架构，约 150 亿参数规模，在高质量输出与合理计算需求之间取得平衡，实现生产级别的视频生成能力。
- Feature 03 / 08
电影级 1080p
支持 720p 与 1080p 双分辨率视频输出，细节清晰、色彩鲜艳、画面具有电影美学，适用于专业内容创作工作流。
- Feature 04 / 08
五大端点
五个专用端点覆盖完整工作流：文生视频、图生视频、参考图生视频、视频编辑（针对性修改）以及视频扩展（无缝延续）。
- Feature 05 / 08
多图参考
同时使用多张参考图引导生成，可锁定整段输出中的角色、场景和视觉风格，确保结果保持高度一致性。
- Feature 06 / 08
灵活时长
可选择 3 到 15 秒之间任意片段长度，默认时长为 5 秒，让创作者精准把控短片、广告和叙事场景的节奏。
- Feature 07 / 08
五种宽高比
原生支持 16:9、9:16、1:1、4:3 和 3:4 五种宽高比，无需裁剪即可覆盖横屏电影、竖屏社交、方形帖子及传统广播格式。
- Feature 08 / 08
极速生成
在单张 H100 GPU 上约 38 秒即可生成一段 1080p 视频，加快迭代周期，为高产量创意流水线带来高性价比的扩展能力。
HappyHorse 1.0是阿里巴巴的旗舰AI视频模型,截至2026年4月在Artificial Analysis榜单上排名第一。它基于单流Transformer架构,约150亿参数,在统一流水线中实现了业界领先的运动稳定性、提示遵循能力和视觉保真度。
HappyHorse提供5个端点,覆盖完整的创作流程:文本生视频、图像生视频、参考生视频(基于主体或风格参考)、提示词驱动的视频编辑,以及视频续写。所有端点均支持720p和1080p输出,时长可在3到15秒之间。
生成的片段时长可从3秒到15秒,默认为5秒。定价按时长线性计算,因此在相同分辨率下,10秒片段的费用正好是5秒片段的两倍。
HappyHorse支持五种宽高比:16:9(横屏)、9:16(竖屏)、1:1(正方形)、4:3(经典横屏)和3:4(经典竖屏)。开箱即用地覆盖电影、社交和传统格式。
HappyHorse采用直观的按秒计费:720p每秒20积分,1080p每秒35积分。总费用等于每秒积分乘以时长。例如,5秒720p片段需100积分,5秒1080p片段需175积分。
截至2026年4月,HappyHorse 1.0在Artificial Analysis榜单的文本生视频和图像生视频两项中均排名第一,领先于Seedance 2.0和Wan 2.7。其单流Transformer架构在运动稳定性和提示遵循方面相比双分支和MoE风格的竞品表现尤为突出。
如何使用 HappyHorse 1.0 文生视频
在 H100 上约 38 秒即可从文本渲染出短电影片段
撰写提示词
用电影化的方式描述场景:主体、动作、镜头运动和光线。具体的视觉细节会带来更清晰、更连贯的画面。