HappyHorse-1.1 is now available through Token Station, giving developers a new video model for both text-to-video and image-to-video generation.
+The important part is not only that there is a new video model. It is that HappyHorse can generate video with native synchronized audio. Instead of producing a silent clip first and then stitching in speech, music, or sound effects later, you can start with a single video generation request that is designed to return motion and sound together.
+That makes HappyHorse-1.1 useful for product demos, social clips, ad concepts, creative agents, and video workflows where iteration speed matters.
+The new model IDs are:
+bailian/happyhorse-1.1-t2v— text to videobailian/happyhorse-1.1-i2v— image to video from a first frame
HappyHorse-1.0 remains available. Token Station exposes both versions so developers can test 1.1 without breaking existing 1.0 workflows.
+Why HappyHorse-1.1 matters
+Most AI video tools still split the job into multiple steps.
+A typical workflow looks like this:
+- Generate a silent video clip.
- Generate or source audio separately.
- Align the audio with the motion.
- Export or post-process the final asset.
That workflow can make sense for polished production. But it is heavy for rapid iteration. If you are testing landing page videos, generating ad concepts, building a creative agent, or producing quick social clips, the extra audio step slows everything down.
+HappyHorse is interesting because synchronized audio is part of the video generation workflow. A single model call can produce a more complete clip: motion plus sound.
+That is especially useful for:
+- product teaser clips
- social media videos
- ad concept drafts
- agent-generated creative assets
- first-frame animation with sound
- quick demos for landing pages or launches
- prototype video generation inside a larger app
The model is not a replacement for every professional editing pipeline. But it gives developers a stronger default output from one API call.
+HappyHorse-1.1 vs 1.0: pricing
+The easiest way to understand the 1.1 update is the price comparison:
+- HappyHorse-1.1 720P:
$0.14/s - HappyHorse-1.0 720P: about
$0.124/s - HappyHorse-1.1 1080P:
$0.18/s - HappyHorse-1.0 1080P: about
$0.221/s
For a 5-second HappyHorse-1.1 clip, that is approximately:
+- 720P:
$0.70 - 1080P:
$0.90
So HappyHorse-1.1 is more expensive at 720P, but cheaper at 1080P.
+That matters if you are building a real product. For low-resolution draft generation, 1.0 may still be attractive. For higher-resolution clips, 1.1's 1080P tier can be a better fit.
+Use HappyHorse-1.1 for text-to-video
+Use bailian/happyhorse-1.1-t2v when you want to generate a video directly from a prompt.
curl -X POST https://models.bytefuture.ai/v1/video/generations \
+ -H "Content-Type: application/json" \
+ -H "Authorization: Bearer APIKEY" \
+ -d '{
+ "model": "bailian/happyhorse-1.1-t2v",
+ "prompt": "A horse galloping across a misty meadow at golden hour, cinematic lighting",
+ "duration": 5,
+ "resolution": "720P"
+ }'For 1080P output, change the resolution:
+{
+ "model": "bailian/happyhorse-1.1-t2v",
+ "prompt": "A cinematic product video of a compact AI coding keypad on a desk",
+ "duration": 5,
+ "resolution": "1080P"
+}Use text-to-video when the prompt is enough to define the scene. It is the fastest path for ideation: product demo concepts, background scenes, social media clips, and quick visual experiments.
+Use HappyHorse-1.1 for image-to-video
+Use bailian/happyhorse-1.1-i2v when you already have a first frame, product shot, character image, or generated still image that you want to animate.
curl -X POST https://models.bytefuture.ai/v1/video/generations \
+ -H "Content-Type: application/json" \
+ -H "Authorization: Bearer APIKEY" \
+ -d '{
+ "model": "bailian/happyhorse-1.1-i2v",
+ "prompt": "The character turns toward the camera and waves, natural motion, warm lighting",
+ "image": "https://example.com/photo.jpg",
+ "duration": 5,
+ "resolution": "1080P"
+ }'Image-to-video is useful when visual consistency matters. Instead of asking the model to invent the whole scene from text, you give it a starting frame and ask it to animate that frame.
+Good use cases include:
+- animating a product shot
- turning a generated image into a short clip
- making a character move while preserving identity
- creating motion variations from the same visual concept
- testing ad creatives from a static mockup
Token Station handles the Bailian-specific first-frame request shape behind the scenes. You pass an image URL through the gateway interface; the gateway converts the request into the format HappyHorse expects upstream.
Why expose both 1.0 and 1.1?
+Versioned model IDs are important for production systems.
+If a provider changes output style, cost, model behavior, or supported features, developers need a way to choose when to move. Replacing 1.0 with 1.1 behind the same name would make existing workflows harder to reproduce and harder to budget.
+That is why Token Station exposes both versions:
+bailian/happyhorse-1.0-t2v
+bailian/happyhorse-1.0-i2v
+bailian/happyhorse-1.1-t2v
+bailian/happyhorse-1.1-i2vYou can keep stable workloads on 1.0, route new experiments to 1.1, and decide which version should become your default after comparing real outputs and costs.
+Why this fits Token Station
+Token Station is designed for model adoption without integration churn.
+When a new model arrives, developers should not need to rebuild their application around another provider-specific API. They should be able to try it behind the same gateway surface, compare it with other models, and decide whether it belongs in production.
+HappyHorse-1.1 is a good example:
+{
+ "model": "bailian/happyhorse-1.1-t2v",
+ "prompt": "A cinematic launch video for a new AI product",
+ "duration": 5,
+ "resolution": "1080P"
+}The gateway handles provider routing, async task creation, polling, result handling, and usage accounting. The developer gets a model ID and a consistent API.
+That is the real benefit: new model launches become something you can test quickly, not a full integration project.
+When should you try HappyHorse-1.1?
+Try HappyHorse-1.1 if you need short video clips with audio and want to move quickly.
+It is a natural fit for:
+- creative agents that generate media assets
- product teams testing demo videos
- marketing teams testing social clips
- ad teams iterating on visual concepts
- apps that turn product images into motion
- workflows where synchronized audio is useful from the first draft
Start with 720P when you are exploring prompts. Move to 1080P when you have a clip direction worth refining.
+Try HappyHorse-1.1
+Start with text-to-video:
+bailian/happyhorse-1.1-t2vOr animate a first frame with image-to-video:
+bailian/happyhorse-1.1-i2vTry it on Token Station:
+ +