How to Use HappyHorse 1.1 API on EvoLink

If you want to use HappyHorse 1.1 through EvoLink, the practical workflow is:

1. Choose the HappyHorse 1.1 model ID that matches your input type.
2. Send a video generation task to POST /v1/videos/generations.
3. Store the returned task ID.
4. Poll GET /v1/tasks/{task_id} or configure callback_url.
5. Download or move the generated video result before your application depends on it.

Quick answer

HappyHorse 1.1 currently has three EvoLink routes:

Do not mix HappyHorse 1.1 with older HappyHorse 1.0 routes. The current HappyHorse 1.1 integration surface on EvoLink is the three-route setup above.

Table of contents

• Confirmed facts
• Choose the right route
• Prerequisites
• Basic request flow
• Text-to-video example
• Image-to-video example
• Reference-to-video example
• Parameters to plan before production
• Async status and callback_url
• Cost planning
• Production rollout checklist
• Troubleshooting
• FAQ

Confirmed facts

This matters because video generation should be treated as an async production job, not a synchronous API call that blocks a web request.

Choose the right route

The biggest integration mistake is choosing a route by name instead of by input asset. Start from what your workflow already has.

Use text-to-video for ideation, image-to-video for asset animation, and reference-to-video when consistency across multiple input images matters.

Prerequisites

Before you make the first request, prepare these items:

For image inputs, use publicly accessible HTTP or HTTPS URLs. Private intranet links, authenticated object storage URLs, and local files are not valid production inputs for the API.

Basic request flow

HappyHorse 1.1 follows EvoLink's video task pattern:

Text-to-video example

curl --request POST \
  --url https://api.evolink.ai/v1/videos/generations \
  --header 'Authorization: Bearer <EVOLINK_API_KEY>' \
  --header 'Content-Type: application/json' \
  --data '{
    "model": "happyhorse-1.1-text-to-video",
    "prompt": "A cinematic product shot of a transparent electric scooter moving through a clean studio space, slow dolly camera, soft reflections",
    "quality": "720p",
    "aspect_ratio": "16:9",
    "duration": 5,
    "callback_url": "https://your-domain.com/webhooks/video-task-completed"
  }'

Text-to-video planning notes

Image-to-video example

curl --request POST \
  --url https://api.evolink.ai/v1/videos/generations \
  --header 'Authorization: Bearer <EVOLINK_API_KEY>' \
  --header 'Content-Type: application/json' \
  --data '{
    "model": "happyhorse-1.1-image-to-video",
    "image_urls": ["https://cdn.example.com/product-hero.png"],
    "prompt": "Animate the product with a smooth camera orbit and subtle studio lighting movement",
    "quality": "720p",
    "duration": 5,
    "callback_url": "https://your-domain.com/webhooks/video-task-completed"
  }'

Image-to-video input rules

Reference-to-video example

curl --request POST \
  --url https://api.evolink.ai/v1/videos/generations \
  --header 'Authorization: Bearer <EVOLINK_API_KEY>' \
  --header 'Content-Type: application/json' \
  --data '{
    "model": "happyhorse-1.1-reference-to-video",
    "prompt": "character1 walks into the scene, picks up the object shown as character2, and places it on the table beside character3",
    "image_urls": [
      "https://cdn.example.com/person.png",
      "https://cdn.example.com/object.png",
      "https://cdn.example.com/scene.png"
    ],
    "quality": "720p",
    "aspect_ratio": "16:9",
    "duration": 5,
    "callback_url": "https://your-domain.com/webhooks/video-task-completed"
  }'

Reference-to-video input rules

Parameters to plan before production

The exact API reference should remain your schema authority, but most production issues come from a small set of decisions.

Keep a server-side allowlist for model IDs. Do not let client input freely choose arbitrary model names.

Async status and callback_url

After creating a task, your application receives a task ID. Store it immediately and query status:

curl --request GET \
  --url https://api.evolink.ai/v1/tasks/<TASK_ID> \
  --header 'Authorization: Bearer <EVOLINK_API_KEY>'

Task responses can include status, progress, model, task information, result details, and usage information. Treat the task ID as the durable link between your user-facing job and the EvoLink generation job.

Polling vs callback_url

Cost planning

For engineering planning, focus on the cost drivers:

For production, log requested duration, actual task status, model ID, quality, and user/account ID. That makes billing review and incident debugging much easier.

Production rollout checklist

Use this checklist before routing real users to HappyHorse 1.1:

This is where EvoLink's unified gateway matters: once your app uses a clean routing layer, switching between HappyHorse, Seedance, Kling, Sora, or another video model is mostly a model selection decision rather than a new vendor integration.

Troubleshooting

Most HappyHorse 1.1 integration issues come from model ID mismatch, invalid image URLs, unsupported route assumptions, or async handling.

If a request example here ever conflicts with the API reference, follow the API reference.

Example implementation pattern

For most products, use a small routing layer instead of hardcoding HappyHorse in every feature.

type HappyHorseRoute =
  | 'text-to-video'
  | 'image-to-video'
  | 'reference-to-video'

const happyHorseModelIdByRoute: Record<HappyHorseRoute, string> = {
  'text-to-video': 'happyhorse-1.1-text-to-video',
  'image-to-video': 'happyhorse-1.1-image-to-video',
  'reference-to-video': 'happyhorse-1.1-reference-to-video',
}

The routing layer should sit behind product logic. Your UI can ask "Do you have a prompt, one first frame, or multiple references?" while your backend turns that answer into a model ID and validated request body.

When to use another EvoLink video model instead

HappyHorse 1.1 is useful for text, first-frame, and image-reference video workflows. It is not the only video route on EvoLink.

FAQ

Which endpoint do I use for HappyHorse 1.1?

What are the HappyHorse 1.1 model IDs?

Which model ID should I use for prompt-only generation?

Which model ID should I use for first-frame animation?

Which model ID should I use for multiple reference images?

Does image-to-video support aspect_ratio?

The image-to-video route derives output aspect ratio from the source image. For explicit aspect ratio planning, use text-to-video or reference-to-video.

Can I use callback_url?

Where do I find current pricing?

Can I use the same EvoLink API key?

Yes. HappyHorse 1.1 uses the same EvoLink API key and billing system as the other EvoLink routes available to your account.

Does HappyHorse 1.1 support video-edit?

The current HappyHorse 1.1 page and API references expose text-to-video, image-to-video, and reference-to-video. Do not use a HappyHorse 1.0 video-edit route as if it were a HappyHorse 1.1 route.

Sources

• HappyHorse 1.1 API page
• HappyHorse 1.1 Text-to-Video API reference
• HappyHorse 1.1 Image-to-Video API reference
• HappyHorse 1.1 Reference-to-Video API reference