All posts
comparisons

HappyHorse 1.1 vs 1.0: Same Prompts, Side by Side, With Sound

We gave Alibaba's HappyHorse 1.1 and HappyHorse 1.0 the same 27 prompts on Flixly and kept the first result each time. HappyHorse 1.1 speaks scripted lines far more clearly and looks more natural, while HappyHorse 1.0 cuts between shots more and acts out story endings. Nine pairs, side by side, with sound.

By Flixly TeamNovember 13, 2026
HappyHorse 1.1 vs 1.0: Same Prompts, Side by Side, With Sound

TL;DR

HappyHorse 1.1 is the better default of Alibaba's two HappyHorse video models. In 27 same-prompt tests on Flixly at 1080p and 10 seconds, a speech recognizer found the scripted line in 5 of 8 HappyHorse 1.1 dialogue clips against 2 of 8 for HappyHorse 1.0, and 1.1's lighting was warmer and its shots steadier. HappyHorse 1.0 cuts between angles more often, acts out story endings, and handles video editing.

HappyHorse 1.1 is the better default: it speaks scripted dialogue far more clearly and holds steadier, more natural-looking shots, while HappyHorse 1.0 still cuts between shots more often and more readily acts out the end of a story. We gave both Alibaba models the same 27 prompts on Flixly, 15 from text and 12 from a photo, all at 1080p and 10 seconds, and kept the first result every time. Nine of those pairs are below, side by side.

Each clip plays twice: the first time with HappyHorse 1.1's sound, the second time with HappyHorse 1.0's. Turn the sound on.

What is HappyHorse 1.1?

HappyHorse is Alibaba's video generation model. HappyHorse 1.0 came out in April 2026, and Alibaba released HappyHorse 1.1 on June 22, 2026. Alibaba describes 1.1 as an upgrade in five areas: smoother motion, more consistent subjects when you give it several reference images, better instruction following, more natural skin and texture, and tighter sync between sound and picture.

Both versions make a clip with its own soundtrack in one pass: dialogue, music and ambient sound come out with the picture. On Flixly both are in the video generator, listed as Happy Horse 1.1 and Happy Horse.

What is the difference between HappyHorse 1.1 and 1.0 on Flixly?

HappyHorse 1.1 HappyHorse 1.0
Released by Alibaba June 2026 April 2026
Text to video Yes Yes
Image to video (first frame) Yes Yes
Reference images Up to 9 1
Video editing No Yes, in Video Effects
Length 3 to 15 seconds 3 to 15 seconds
Resolution 720p and 1080p 720p and 1080p
Aspect ratios 16:9, 9:16, 1:1, 4:3, 3:4, 4:5, 5:4, 21:9, 9:21 16:9, 9:16, 1:1, 4:3, 3:4
Sound Generated with the video Generated with the video
Median time in our test, text to video 175 seconds 167 seconds
Median time in our test, image to video 130 seconds 171 seconds

Times are wall-clock on Flixly from pressing generate to a finished 10-second 1080p clip, so they include queueing.

How we tested

  • Same prompt, same settings. Every pair used one prompt, 1080p, 10 seconds and 16:9. Image-to-video pairs started from the same still, made the same day with Qwen Image 2.1.
  • First result only. No rerolls, no editing. The only changes here are the side-by-side layout and compression for the web.
  • 27 pairs. 15 text-to-video prompts aimed at dialogue, sound and motion, and 12 image-to-video prompts. Eight of the prompts contain a line of dialogue in quotes.
  • A rough dialogue check. We ran the eight dialogue clips from each model through an automatic speech recognizer to see whether the scripted line was actually spoken.

Which one speaks dialogue better?

HappyHorse 1.1, clearly. The speech recognizer picked out the scripted line in 5 of the 8 HappyHorse 1.1 clips and 2 of the 8 HappyHorse 1.0 clips. It is a rough check, not a lip-sync score, but it matches what you hear: 1.1 tends to say the words you wrote, while 1.0 more often fills the scene with sound and a line that does not quite match.

Prompt: "A pilot in a sci-fi cockpit says over the radio: 'Control, we are clear for jump in three, two, one.'" Text to video, 1080p, 10 seconds, 16:9. HappyHorse 1.1 took 141 seconds, HappyHorse 1.0 took 164 seconds.

This pair also shows the trade-off. HappyHorse 1.1 says the countdown and stays in the cockpit. HappyHorse 1.0 delivers a less clear line but jumps to hyperspace at "one", acting out what the countdown is for.

Prompt: "A barista calls out an order: 'One oat flat white for Sam!' in a busy coffee shop" Text to video, 1080p, 10 seconds, 16:9. HappyHorse 1.1 took 175 seconds, HappyHorse 1.0 took 166 seconds.

HappyHorse 1.1 built a small sequence: the pour, a close-up of the latte art forming, then the barista handing over the cup. The speech recognizer caught "for Sam" in its audio and nothing recognizable in 1.0's.

Prompt: "A cooking show host chops garlic and says: 'The secret is to never rush the garlic.'" Text to video, 1080p, 10 seconds, 16:9. HappyHorse 1.1 took 246 seconds, HappyHorse 1.0 took 168 seconds.

Both models got this line across. The difference is in the close-up: HappyHorse 1.0's garlic comes out as smooth, egg-like discs, while 1.1's looks like garlic being minced.

Which one looks more natural?

HappyHorse 1.1, most of the time. Alibaba says 1.1 cuts down the oily, over-sharpened skin that gives AI faces away, and in our pairs its lighting was usually warmer and more even. In several pairs HappyHorse 1.0 chose darker, moodier lighting for the same prompt.

Prompt: "A grandmother tells a story to two children by a fireplace and says: 'And that is how the river got its name.'" Text to video, 1080p, 10 seconds, 16:9. HappyHorse 1.1 took 201 seconds, HappyHorse 1.0 took 171 seconds.

Which one handles motion and camera work better?

They make different choices. HappyHorse 1.1 tends to commit to one shot and hold it, and HappyHorse 1.0 cuts between shots more often, the way an editor would.

Prompt: "A horse gallops across a misty meadow at dawn, hooves thudding, low angle camera" Text to video, 1080p, 10 seconds, 16:9. HappyHorse 1.1 took 145 seconds, HappyHorse 1.0 took 146 seconds.

HappyHorse 1.1 tracked the horse side-on for the full ten seconds with a steady gait. HappyHorse 1.0 made three shots: hooves in the grass, a side view, and the horse running off into the sunrise.

Prompt: "Waves crash against a lighthouse during a storm at night, the beam sweeping through the rain" Text to video, 1080p, 10 seconds, 16:9. HappyHorse 1.1 took 122 seconds, HappyHorse 1.0 took 167 seconds.

This one is close. HappyHorse 1.1 added lightning and a huge wave; HappyHorse 1.0 made the sweeping beam, which the prompt asked for, much more visible.

How do they compare from a photo?

In image to video the two stay much closer, because both keep the photo's look. The differences show up in how literally they read the prompt and how much they move.

Prompt: "The baker smiles, dusts flour off her hands and says: 'Fresh bread, straight from the oven!'" Image to video from a Qwen Image 2.1 still, 1080p, 10 seconds. HappyHorse 1.1 took 130 seconds, HappyHorse 1.0 took 373 seconds.

Both models said the line, kept the "PAIN DE SEIGLE" chalkboard sharp, and kept the man from the photo even though our prompt said "her". The photo wins over the prompt in both.

Prompt: "She tilts the umbrella back, looks up at the rain and smiles as cars pass behind her" Image to video from a Qwen Image 2.1 still, 1080p, 10 seconds. HappyHorse 1.1 took 128 seconds, HappyHorse 1.0 took 184 seconds.

Our prompt mentioned an umbrella the photo does not have. HappyHorse 1.1 followed the prompt and pulled a black umbrella into the shot. HappyHorse 1.0 stuck to the photo. If your prompt and your image disagree, 1.1 is more likely to change the picture to match the words.

Prompt: "The astronaut turns slowly toward the two moons as dust blows across the red sand" Image to video from a Qwen Image 2.1 still, 1080p, 10 seconds. HappyHorse 1.1 took 128 seconds, HappyHorse 1.0 took 170 seconds.

HappyHorse 1.1 turned the astronaut and moved the camera around to a profile against the moons. HappyHorse 1.0 kept the shot almost still, with dust drifting past.

When is HappyHorse 1.0 still the better choice?

  • When you want editing in the generation. In our tests 1.0 cut between angles more often, which can save you assembling shots yourself.
  • When the story matters more than the line. 1.0 was more likely to show the payoff of a prompt, like the hyperspace jump.
  • When you want to edit an existing video. Video editing on Flixly uses HappyHorse 1.0 in Video Effects; 1.1 does not edit videos.

For everything else, especially anything with dialogue, start with HappyHorse 1.1.

How to use HappyHorse 1.1 on Flixly

  1. Open the video generator with HappyHorse 1.1 selected.
  2. Write what happens and, in quotes, anything a character should say: A news anchor says: "Good evening."
  3. Pick a length from 3 to 15 seconds, 720p or 1080p, and an aspect ratio.
  4. To animate a photo, switch to image to video and upload it. Make sure the prompt does not describe things the photo lacks unless you want them added.
  5. Check the price shown on the button, then generate. A 10-second 1080p clip took us about two to three minutes.

Which one should you use?

If you want Use
A character to say a specific line HappyHorse 1.1
Natural faces and even lighting HappyHorse 1.1
One steady, continuous shot HappyHorse 1.1
Vertical or ultra-wide formats (4:5, 21:9, 9:21) HappyHorse 1.1
Several reference images of a character or product HappyHorse 1.1
A clip that cuts between angles on its own HappyHorse 1.0
Editing an existing video HappyHorse 1.0

Model pages: HappyHorse 1.1 and HappyHorse 1.0. For another Alibaba pair, see Wan 3.0 vs Wan 3.0 Prime.

Frequently Asked Questions

Is HappyHorse 1.1 better than HappyHorse 1.0?▾

For most clips, yes. In Flixly's test of Alibaba's two models with the same 27 prompts, HappyHorse 1.1 spoke scripted dialogue more clearly (the line was recognized in 5 of 8 clips against 2 of 8 for HappyHorse 1.0) and its lighting was warmer and more even. HappyHorse 1.0 cut between angles more often and acted out story endings more readily.

What changed in HappyHorse 1.1?▾

Alibaba released HappyHorse 1.1 on June 22, 2026, with smoother motion, more consistent subjects across several reference images, better instruction following, more natural skin and texture, and tighter sound-picture sync. On Flixly, HappyHorse 1.1 also adds four aspect ratios (4:5, 5:4, 21:9 and 9:21) and takes up to 9 reference images, against 1 for HappyHorse 1.0.

Do HappyHorse 1.1 and 1.0 generate sound and dialogue?▾

Yes. Both of Alibaba's HappyHorse models generate the soundtrack together with the picture, including dialogue, music and ambient sound, and every clip in Flixly's test came back with audio. Put spoken lines in quotes in the prompt. HappyHorse 1.1 said the written line far more reliably than HappyHorse 1.0 in Flixly's comparison.

Can HappyHorse 1.1 edit an existing video?▾

No. Video editing on Flixly uses HappyHorse 1.0, which Alibaba offers with a video-editing mode: upload a clip in Video Effects and describe the change. HappyHorse 1.1 on Flixly covers text to video, image to video from a first frame, and reference to video with up to 9 images.

How long and how sharp can HappyHorse videos be?▾

Both HappyHorse 1.1 and HappyHorse 1.0 from Alibaba make clips from 3 to 15 seconds at 720p or 1080p on Flixly, at 24 frames per second. HappyHorse 1.1 supports nine aspect ratios, including 4:5, 21:9 and 9:21, while HappyHorse 1.0 supports five: 16:9, 9:16, 1:1, 4:3 and 3:4.

How long does HappyHorse 1.1 take to generate a video?▾

In Flixly's test, a 10-second 1080p clip from Alibaba's HappyHorse 1.1 took a median of 175 seconds from text and 130 seconds from a photo, measured from pressing generate to the finished file. HappyHorse 1.0 took a median of 167 seconds from text and 171 seconds from a photo.

Tools mentioned in this post

HappyHorse 1.1HappyHorseHappy HorseAlibabaAI Video GeneratorText to VideoImage to VideoAI Video With SoundModel ComparisonFlixly

Related Articles

More on comparisons from the Flixly blog

Wan 3.0 vs HappyHorse 1.1: 20 Prompts on Four Alibaba Video Models
comparisons

Wan 3.0 vs HappyHorse 1.1: 20 Prompts on Four Alibaba Video Models

We gave four Alibaba video models, Wan 3.0, Wan 3.0 Prime, HappyHorse 1.1 and HappyHorse 1.0, the same 20 prompts on Flixly and kept the first result each time. On these prompts Wan stayed in one continuous take and followed camera directions, while HappyHorse cut between angles. Wan 3.0 Prime was the fastest by far.

Wan 3.0 vs Wan 3.0 Prime: Same Prompts, Side by Side, Timed
Model Launches

Wan 3.0 vs Wan 3.0 Prime: Same Prompts, Side by Side, Timed

We gave Alibaba's Wan 3.0 and Wan 3.0 Prime the same prompts on Flixly, from 5-second 480p drafts to 30-second 1080p single takes, and kept the first result each time. The clips looked equally good, and Wan 3.0 Prime finished first in all 22 pairs, usually in about two minutes. Five pairs, side by side, with sound.

LTX 2.5 Fast and Pro Are Live on Flixly: Video and Sound in One Pass
Model Launches

LTX 2.5 Fast and Pro Are Live on Flixly: Video and Sound in One Pass

Lightricks' LTX 2.5 is live on Flixly in two tiers. LTX 2.5 Fast makes 6 to 20 seconds of video with dialogue, sound effects and ambience at up to 1440p, from text, a start image or your own audio. LTX 2.5 Pro adds sharper faces, textures and on-screen text. See real results with their settings.

Seedance 2.5 Flash Is Live on Flixly: Faster 30-Second AI Video
Model Launches

Seedance 2.5 Flash Is Live on Flixly: Faster 30-Second AI Video

Seedance 2.5 Flash, ByteDance's faster version of Seedance 2.5, is live on Flixly. Make up to 30 seconds of video at 480p or 720p with synchronized audio from a prompt, a start and end frame, or up to 10 images, 5 videos and 5 audio clips, for about a third less per second. See four real results with their prompts.

Ready to create with comparisons?

Jump straight into Flixly's AI studio and try comparisons with 193+ models — free to start.