MiniMax MCP
Generate MiniMax-H3 videos (768P/2K) through MCP; after submission, query the final status and video URL by task ID.
Minimax
Generate videos from text, images, or audio guidance.
Try the API
API Key
Please obtain an API Key first
This is only a basic call example. See the full docs for more parameters and advanced usage.
View DocsMiniMax H3 · Multimodal Video Generation API
MiniMax H3 Video Generation
Generate 4–15 second videos from text, up to 9 reference images, or up to 3 reference audio clips with a single REST API. Unified asynchronous tasks, callbacks, billing, and CDN delivery integrate directly into your content product.
Text-to-video / Image-to-video / Audio-guided768P / 2KAsynchronous Tasks + Webhook
🎬
3
Video Generation Modes
⏱️
4-15s
Available Video Duration
🖼️
1-9
Reference Images
🎵
1-3
Reference Audio Clips
One API, Three Creative Starting Points
No action or multiple paths required; the API automatically determines the generation mode based on the content assets.
✍️
Text-to-Video
Generate short videos directly from descriptions of scenes, actions, camera shots, and styles.
🖼️
First-and-Last-Frame Image-to-Video
Control the opening and ending visuals with first_frame and last_frame.
🎵
Multimodal References
Combine reference images, videos, and audio to guide subjects, actions, sound, and rhythm.
📡
Asynchronous Delivery
Receive a task ID immediately, then get the final CDN video through polling or callback_url.
curl
curl -X POST 'https://api.acedata.cloud/minimax/videos' \
-H 'Authorization: Bearer YOUR_API_KEY' \
-H 'Content-Type: application/json' \
-d '{
"model": "MiniMax-H3",
"content": [{
"type": "text",
"text": "A red fox running through a snowy forest at dawn, cinematic tracking shot"
}],
"resolution": "2K",
"ratio": "16:9",
"duration": 5
}'Start Generating with Just a Few Lines of Code
Parameters directly express creative intent, while the platform handles long-running task execution, result storage, and precise billing.
1
Build content
All modes must provide text; images, videos, and audio indicate their purpose through their corresponding types and roles.
2
Submit an Asynchronous Task
The creation API is always asynchronous and returns a task_id immediately upon success.
3
Get Permanent Results
Retrieve the Ace Data Cloud CDN video from task.content.url via /minimax/tasks or a callback.
Typical scenarios from source material to finished video
Three real API samples cover product activation, character storytelling, and audio-visual reference workflows.
Brand and e-commerce shorts
Create commercial shorts from product visuals and shot descriptions, ideal for product showcases, launch assets, and social media ads.
MiniMax-H3 · 16:9 · First-frame image-to-video
Consistent character storytelling
Use multiple proprietary reference images to improve consistency in characters, clothing, props, and scene styles, ideal for continuous shots and serialized content.
MiniMax-H3 · 16:9 · Multi-image reference
Audio-visual rhythm shorts
Combine proprietary motion video and audio references to guide dance movements, camera motion, and visual rhythm, ideal for vertical content creation.
MiniMax-H3 · 9:16 · Video and audio referenceGo live in three steps
Reuse Ace Data Cloud's unified authentication and task system without building a separate media pipeline.
01
Get a Token
Create an application in the console and obtain a unified Bearer Token.
02
Submit generation
Choose the duration and aspect ratio, then add text, image, or audio assets.
03
Deliver video
Poll the task or receive a callback, then integrate the CDN video into your product workflow.
Unified Video API for Production
Consolidate task execution, error semantics, usage records, and file delivery into a stable contract.
| Capability | Ace Data Cloud MiniMax H3 | Some Others |
|---|---|---|
| Multimodal input | Unified entry point for text, images, video, and audio | Multiple APIs to assemble yourself |
| Long-running tasks | Polling and Webhook | Need to build your own task queue |
| Result files | Automatically stored on the platform CDN | Inconsistent link lifecycles |
| Failure billing | No charge for failures | Need to reconcile bills yourself |
Select a workflow by content
The model is fixed as MiniMax-H3, and the mode is automatically inferred from the official V2 multimodal content array.
TEXT
text
Pass text content only to generate visuals directly from ideas, scripts, and shot descriptions.
FRAME
first_frame / last_frame
Use the first frame or the first and last frames to control the video's opening and ending visuals.
REFERENCE
reference media
Combine reference images, video, and audio to control the subject, motion, sound, and rhythm.
TASK
task_id / callback
Create asynchronous tasks and retrieve results through task queries or callbacks.
Transparent resolution pricing
Billed by video seconds, with charges only for successful tasks.
768P
$0.057143 / second
- As low as $0.228572 for 4 seconds
- Ideal for rapid creative validation
- Three input modes
2K
$0.091429 / second
- As low as $0.365716 for 4 seconds
- Default output resolution
- Asynchronous task queries
Frequently Asked Questions
Key notes on modes, assets, duration, and task execution.
Do I need to pass action?
No. When content only has text, it is text-to-video; first_frame / last_frame are image-to-video; reference_* are multimodal reference-to-video. All modes require non-empty text.
What aspect ratios and durations are supported?
Resolution is required, with support for 768P and 2K; supported ratios include adaptive, 21:9, 16:9, 4:3, 1:1, 3:4, and 9:16; duration is an integer from 4–15 seconds.
How do I get the results?
The creation API is always asynchronous and immediately returns task_id; use /minimax/tasks to query it. Upon success, get the video from task.content.url, or provide callback_url.
Are failed tasks charged?
No. Only tasks that successfully complete and return video results are billed based on the final duration.
More Public Video Services
Combine other video capabilities based on model style and workflow needs.
Integrate MiniMax H3 into Your Product
One Token, one generation endpoint, one task system—start building multimodal video experiences.
Beta MiniMax H3 Videos API
POSThttps://api.acedata.cloud/minimax/videos
Generate videos with Minimax multimodal video models using synchronous waits, asynchronous tasks, or completion callbacks.
MiniMax H3 Tasks API
POSThttps://api.acedata.cloud/minimax/tasks
Retrieve one or multiple Minimax video tasks with processing status, final videos, and standard task metadata.
Beta 提示词增强
POSThttps://api.acedata.cloud/minimax/prompt-enhancement
Minimax Prompt Enhancement generation with asynchronous task delivery.
Beta 视频重生成
POSThttps://api.acedata.cloud/minimax/regenerate
Minimax Video Regeneration generation with asynchronous task delivery.
Packages
We offer various credit packages. The more you purchase, the higher the discount you receive.
up to46% OFFPrice
MiniMax H3 Videos API
MiniMax H3 Tasks API
提示词增强
视频重生成
No Pricing Information
Agent app
Installation Scope
Operating System
Configuration Method
Connect the current service in Claude Code:
- Current project: Merge the
mcpServersservice entry below into.mcp.jsonin the project root directory, retaining the existing configuration. - Current user: Run the command below in the terminal; the configuration can be reused across projects.
- Project configuration requires setting
ACEDATACLOUD_API_TOKENin the terminal: useexport ACEDATACLOUD_API_TOKEN='YOUR_API_TOKEN'on macOS/Linux, and$env:ACEDATACLOUD_API_TOKEN = 'YOUR_API_TOKEN'in Windows PowerShell; replaceYOUR_API_TOKENwith your AceDataCloud API Token. - Run
claudein the same terminal where this variable has been set; new terminals or desktop-launched clients must also be able to read this variable. - After entering the Claude Code session, run
/mcpto confirm that the service is connected and the tools have been loaded; when using the project configuration for the first time, check and approve the service as prompted.
Claude Code
export ACEDATACLOUD_API_TOKEN='YOUR_API_TOKEN'
claude mcp add --scope user --transport http 'minimax-mcp' 'https://minimax.mcp.acedata.cloud/mcp' --header 'Authorization: Bearer ${ACEDATACLOUD_API_TOKEN}'
MiniMax Video
Provide AI Agent with MiniMax H3 text-to-video, multi-image references, audio guidance, and task polling instructions.
Agent app
Installation Scope
Install the current Skill to Claude Code. Project-level configurations are shared across the repository, while user-level configurations apply to all projects on this machine.
Claude Code
npx skills add acedatacloud/skills --skill 'minimax-video' -a 'claude-code' --yes
Introduction
APIs
Pricing
MCP
Skill