MiniMax H3 API

Generation waits for completion by default. Send async=true or callback_url to receive a task_id, then poll /minimax/tasks. The published model is MiniMax-H3; H3 Max, prompt enhancement and regeneration are not published.

Generate 4–15 second videos from text, first/last frames, and multimodal references through AceDataCloud.

MCP integration: MiniMax H3 MCP.

The MCP supports MiniMax-H3 at 768P or 2K, using text, reference images, or audio. After generation returns a task ID, keep querying the task until its status is succeeded, failed, or cancelled. Only use task.content.url after succeeded; for failure or cancellation, inspect the task's error information. Receiving a task ID does not mean the video is ready. See the task retrieval procedure.

curl -X POST https://api.acedata.cloud/minimax/videos \
  -H "Authorization: Bearer $ACEDATACLOUD_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "MiniMax-H3",
    "content": [
      {
        "type": "text",
        "text": "A red fox running through a snowy forest at dawn, low tracking shot"
      }
    ],
    "resolution": "2K",
    "ratio": "16:9",
    "duration": 4,
    "async": true
  }'

Poll the returned task_id:

curl -X POST https://api.acedata.cloud/minimax/tasks \
  -H "Authorization: Bearer $ACEDATACLOUD_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"action":"retrieve","id":"TASK_ID"}'

The API uses a content array for the required text prompt and optional image, video, and audio references. Poll the task about every 10 seconds until it reaches a terminal status.

• Video generation guide
• Task API guide
• AceDataCloud console