In-depth architectural comparison of the Minimax H3 and Magic Hour MCP Server MCP servers. Compare execution transports, security boundaries, tool capabilities, quality scores, and ready-to-paste client installation snippets for Claude, Cursor, Windsurf, and VS Code.
At a Glance & Executive Verdict
Minimax H3
Multimedia Process · Local stdio
Quality: 45/100 (Fair) | Auth: No auth required
Magic Hour MCP Server
Multimedia Process · Remote HTTP/SSE
Quality: 60/100 (Good) | Auth: No auth required
Verdict Summary: Choose Minimax H3 if you need specialized Multimedia Process tools running via a local process. Choose Magic Hour MCP Server if your workspace requires Multimedia Process integration with remote web transport. Both servers can be configured concurrently in your client's mcpServers manifest.
Which MCP Server Should You Choose?
M
Choose Minimax H3 when:
You need dedicated capabilities in the Multimedia Process domain.
You prefer local stdio subprocess transport architecture.
Your security boundary fits: No auth required (Free / Open Source).
MiniMax H3 is a multimodal AI video generator supporting text‑to‑video and first‑last‑frame control. It uses combined image, video and audio references to lock characters, motions and styles. It delivers native 2K videos with synchronized stereo audio in multiple aspect ratios for ads, e‑commerce, film previsualization, game concepts and social content prototyping.
Create and edit AI videos, images, GIFs, and audio through 44 MCP tools.
No explicit tool names declared in metadata yet. Check project README on main listing page.
Magic Hour MCP Server Tools (45)
ping
Check that the Magic Hour MCP server is reachable.
wait_for_video_project
Poll a video project until it completes, errors, is canceled, or times out. Returns sanitized download fields. Use `exact_download_urls[n]` or `downloads[n].url` exactly as returned; do not shorten it, remove query parameters, or append expiration metadata.
wait_for_image_project
Poll an image project until it completes, errors, is canceled, or times out. Returns the final project JSON and, when complete, attempts to inline image downloads for Inspector or compatible clients. Returns sanitized download fields. Use `exact_download_urls[n]` or `downloads[n].url` exactly as returned; do not shorten it, remove query parameters, or append expiration metadata.
wait_for_audio_project
Poll an audio project until it completes, errors, is canceled, or times out. Returns the final project JSON and, when complete, attempts to inline audio downloads for Inspector or compatible clients. Returns sanitized download fields. Use `exact_download_urls[n]` or `downloads[n].url` exactly as returned; do not shorten it, remove query parameters, or append expiration metadata.
Ready-to-Paste Client Configurations
Paste either (or both) of these JSON server blocks into your client config file (e.g. claude_desktop_config.json or ~/.cursor/mcp.json).
Minimax H3 is categorized under Multimedia Process and uses a local stdio subprocess. In contrast, Magic Hour MCP Server belongs to Multimedia Process using remote streaming HTTP/SSE transport. Select Minimax H3 when you need capabilities focused on multimedia process and Magic Hour MCP Server when you require tools for multimedia process.
Fetch a image `downloads[n].url` from a completed image project and return it as inline MCP image content for compatible clients. Pass the exact full signed URL from `downloads[n].url` without trimming query parameters; `expires_at` is separate metadata, not part of the URL.
fetch_audio_download
Fetch a audio `downloads[n].url` from a completed audio project and return it as inline MCP audio content for compatible clients. Pass the exact full signed URL from `downloads[n].url` without trimming query parameters; `expires_at` is separate metadata, not part of the URL.
fetch_video_download
Fetch a video `downloads[n].url` from a completed video project and return it as an embedded MCP binary resource for compatible clients. Pass the exact full signed URL from `downloads[n].url` without trimming query parameters; `expires_at` is separate metadata, not part of the URL.
account_retrieve
Get the current credit balance and subscription details of the account that owns the API key.
saved_items_list
Returns active saved items owned by the authenticated account, newest first. Each item includes every saved asset with a durable file_path for reuse in compatible generation APIs and a temporary signed URL for previewing or downloading. Filter by type to find characters, references, voices, moodboards, or brand kits. To fetch the next page, pass the response's next_cursor as cursor.
video_assets_generate_presigned_url
Generates a list of pre-signed upload URLs for the assets required. This API is only necessary if you want to upload to Magic Hour's storage. Refer to the [Input Files Guide](https://docs.magichour.ai/integration/inputs-and-outputs) for more details.
The response array will match the order of items in the request body.
**Valid file extensions per asset type**:
- video: mp4, m4v, mov, webm
- audio: mp3, wav, aac, flac, webm, weba, m4a, opus, ogg, oga, aiff, amr
- image: png, jpg, jpeg, jfif, heic, heif, webp, avif, jp2, tiff, tif, bmp
- gif: gif, webp, webm
> Note: `gif` is only supported for face swap API `video_file_path` field.
Once you receive an upload URL, send a `PUT` request to upload the file directly.
Example:
```
curl -X PUT --data '@/path/to/file/video.mp4' \
https://videos.magichour.ai/api-assets/id/video.mp4?<auth params from the API response>
```
MCP guidance:
- This only creates presigned upload URLs. For local files, upload the raw bytes to each returned `upload_url` outside the generation call, then pass the matching `file_path` into the create tool.
- For `*_file_path` values, prefer an existing Magic Hour file path or a `file_path` returned by the upload-URL endpoint after the file bytes are uploaded. Direct public media URLs may work when they are stable, fetchable, and return raw file bytes, but hotlinked URLs can fail; when in doubt, use the presigned upload flow first and pass the returned `file_path`.
face_detection_retrieve_details
Get the details of a face detection task.
Use this API to get the list of faces detected in the image or video to use in the [face swap photo](https://docs.magichour.ai/api-reference/image-projects/face-swap-photo) or [face swap video](https://docs.magichour.ai/api-reference/video-projects/face-swap-video) API calls for multi-face swaps.
face_detection_detect_faces
Detect faces in an image or video.
Use this API to get the list of faces detected in the image or video to use in the [face swap photo](https://docs.magichour.ai/api-reference/image-projects/face-swap-photo) or [face swap video](https://docs.magichour.ai/api-reference/video-projects/face-swap-video) API calls for multi-face swaps.
Note: Face detection is free to use for the near future. Pricing may change in the future.
MCP guidance:
- This starts an async face-detection task and returns an `id`. Use the face-detection details endpoint with that id to retrieve detected faces before doing individual face swaps.
- For `*_file_path` values, prefer an existing Magic Hour file path or a `file_path` returned by the upload-URL endpoint after the file bytes are uploaded. Direct public media URLs may work when they are stable, fetchable, and return raw file bytes, but hotlinked URLs can fail; when in doubt, use the presigned upload flow first and pass the returned `file_path`.