---
title: "RunAPI 経由で Hermes Agent に Hailuo を使う — 動画 API ガイド"
url: "https://runapi.ai/ja/hermes-hailuo.md"
canonical: "https://runapi.ai/ja/hermes-hailuo"
locale: "ja"
model: "hailuo"
---

# Hermes Agent で Hailuo を使う。

Hailuo は MiniMax の動画生成モデルで、正確な物理シミュレーションを伴うネイティブ 1080p 映像を生成し、6 バリアントを揃えています —— Hailuo 02 と 2.3 で、それぞれ pro と standard ティアにテキスト→動画と画像→動画のエンドポイントを備えています。Hermes Agent は custom:runapi provider を使って RunAPI 経由で呼び出します —— チャット用に設定したものと同じキーと base URL を使います。

## API example

```bash
curl -X POST https://runapi.ai/api/v1/hailuo/text_to_video \
  -H "Authorization: Bearer $RUNAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "hailuo-02-text-to-video-pro",
    "prompt": "A glass marble rolls across a wooden desk, catches the edge, and falls in slow motion. Camera follows the marble. Soft window light, shallow depth of field."
  }'

```

### Response

```json
{
  "task_id": "tsk_abc123",
  "status": "pending",
  "model": "hailuo-02-text-to-video-pro"
}

```

## How it works

1. **Configure RunAPI** — Set the RUNAPI_API_KEY environment variable. If you already added RunAPI as a custom:runapi provider in Hermes Agent for chat, the same key and base_url work for video generation — no extra configuration needed.
2. **Call Hailuo text_to_video or image_to_video** — Send a POST request to /api/v1/hailuo/text_to_video with model set to hailuo-02-text-to-video-pro and a scene-describing prompt. For image-anchored video, use the image_to_video endpoint with first_frame_image_url to preserve your source subject.
3. **Poll for the result** — The endpoint returns a task_id immediately. Poll the task status endpoint until the status changes to completed, then retrieve the generated video URL from the response. Typical turnaround is under two minutes for pro tier.

## Parameters

| Parameter | Type | Description |
|-----------|------|-------------|
| `model` | `string` | Required. One of: hailuo-02-text-to-video-pro, hailuo-02-text-to-video-standard, hailuo-02-image-to-video-pro, hailuo-02-image-to-video-standard, hailuo-2.3-image-to-video-pro, hailuo-2.3-image-to-video-standard. |
| `prompt` | `string` | Required. Text description of the video scene, camera movement, and lighting. |
| `first_frame_image_url` | `string` | Required for image_to_video endpoints. URL of the source image used as the first frame. |
| `last_frame_image_url` | `string` | Optional. URL of an image used as the last frame. Available on Hailuo 02 image-to-video only. |
| `duration_seconds` | `integer` | Optional. Video length in seconds. Accepted values: 6 or 10. Defaults to 10 on standard, fixed on pro. |
| `output_resolution` | `string` | Optional. Output resolution. 512p or 768p for 02 standard; 768p or 1080p for 2.3 variants. |
| `callback_url` | `string` | Optional. Webhook URL that receives a POST when the task completes. |

## FAQ

### Can I use Hailuo in Hermes Agent?

Yes. Configure RunAPI as a custom:runapi provider in Hermes Agent, then call the text_to_video or image_to_video endpoint with a Hailuo model slug. The same RUNAPI_API_KEY you use for chat handles video generation.

### What is the difference between Hailuo 02 and Hailuo 2.3?

Hailuo 02 supports both text-to-video and image-to-video with first-frame and last-frame control. Hailuo 2.3 is image-to-video only but adds improved body motion, micro-expressions, and dedicated anime and illustration rendering styles.

### How does duration-based pricing work on standard tiers?

Standard tier pricing varies by duration_seconds. For text-to-video standard, a 6-second clip costs 30 cents and a 10-second clip costs 50 cents. Image-to-video standard pricing also factors in output_resolution — a 6s 512p clip starts at 12 cents.

### Does Hermes Agent handle the async polling automatically?

No. Hailuo is asynchronous — the API returns a task_id immediately. Your Hermes Agent workflow needs to poll the task status endpoint or provide a callback_url. Typical generation time is 60-120 seconds.

### Can I control both the first and last frame of a Hailuo video?

Only on Hailuo 02 image-to-video endpoints. Pass first_frame_image_url and last_frame_image_url together to define start and end frames. Hailuo 2.3 image-to-video accepts first_frame_image_url only.

### How do I get the best results from Hailuo without wasting credits?

Start with the standard tier (lower cost) and shorter durations to test your prompt. Once you find the right prompt and composition, switch to pro tier for the final generation. Use negative prompts to reduce artifacts. The standard tier is also good for A/B testing different creative directions quickly.

### Can Hermes Agent extend or chain Hailuo videos in a multi-step workflow?

Yes. Hermes Agent can generate an initial clip, then pass it back to Hailuo for extension, or chain it with ElevenLabs for voiceover, building a complete video sequence in one workflow run.


## Links

- [Hermes Agent セットアップガイド →](https://runapi.ai/ja/hermes-agent)
- [Hailuo モデル →](https://runapi.ai/ja/models/hailuo)
- [Model catalog](https://runapi.ai/ja/models)
- [API docs](https://runapi.ai/ja/docs)