通用的全模态生成模型,支持对文本、图像、视频、声音组成的多模态上下文的统一理解能力、能够输出具备原生双声道的音视频,最高可支持 15s 2K 分辨率。MiniMax H3 具备商用级的多场景内容生成能力,在指令遵循、文字与品牌信息呈现、V2V Motion Transfer(视频到视频动作迁移)等方面表现出色,可实现精准、可控的多模态内容编辑与生成,广泛适用于广告、品牌、电商、产品设计、UI/UX、游戏等商业场景。
| 720P | ¥0.500 / per second |
| 1080P | ¥0.800 / per second |
| Input video 720P | ¥0.500 / sec |
| Input video 1080P | ¥0.800 / sec |
Every route uses the price above. Routes are not priced per token.
| Provider | Latency | Uptime | Status |
|---|---|---|---|
minimax MiniMax-H3 | 84ms | 100.00% |
import os
import time
import requests
API_BASE = "https://api.openroute.cn/api/minimax"
headers = {
"Authorization": f"Bearer {os.environ['ONTOROUTE_API_KEY']}",
"Content-Type": "application/json",
}
created = requests.post(
f"{API_BASE}/v2/video_generation",
headers=headers,
json={
"model": "minimax-h3",
"content": [
{"type": "text", "text": "一只可爱的猫在窗台晒太阳,镜头缓慢拉近"},
],
"duration": 5,
"resolution": "768P",
},
)
created.raise_for_status()
task_id = created.json()["task_id"]
while True:
query = requests.get(f"{API_BASE}/v2/query/video_generation/{task_id}", headers=headers)
query.raise_for_status()
data = query.json()
if data.get("status") == "Success":
print(data["content"]["url"])
break
if data.get("status") == "Fail":
raise RuntimeError(data)
time.sleep(10)暂无调用数据
| Input images | ¥0.200 / image |
| Free images per task | 5 images |
Video is billed by duration. Unlisted resolutions use the fallback price. Duration is rounded up to the billing unit. Input materials are billed separately: images per image (within the per-task free allowance are not charged), input video per second (falls back to the output price when no dedicated tier is listed).
This model does not publish tunable parameters.