MiniMax H3 license: AI-generation identifier is encouraged
MiniMax H3 license III.3.b encourages, not requires, an AI-generation identifier on files. Sume documents a visible plate or burned captions, not metadata.

No, it is not required. Section III.3.b of the MiniMax H3 Community License says you are "encouraged to" add an AI-generation identifier to files produced using generative AI models including MiniMax H3. The same list encourages a "Powered by MiniMax H3" notice and a public write-up. The text does not say what the identifier must be.
This quotes the license, read 2026-10-01; it is not legal advice. What Sume documents is a visible plate (Timeline compose) and burned-in text (video captions). It does not document embedded metadata or a provenance feature.
What does the license ask for, and what does it leave open?
The wording is "an AI-generation identifier", with no format. Exhibit A, item 12, separately forbids publishing machine-generated content to a public environment without clearly and prominently disclosing it. That item is a use restriction, not an encouragement.
| Clause | Wording | Strength |
|---|---|---|
| III.3.b | Add an AI-generation identifier to files from generative AI models | Encouraged |
| III.3.a | "Powered by MiniMax H3" notice | Encouraged |
| Exhibit A, 12 | Disclose machine-generated content in public environments | Use restriction |
What can Sume put on the video itself?
Two documented routes. Timeline compose puts one still and one video on screen in one MP4; in overlay mode the plate is placed by position, width_ratio and margin_ratio. Video captions can burn authored overlay text from cues or segments, which skips speech-to-text.
How do I burn a text label with captions?
Send phrase-level cues with text, start and end in seconds. The docs say this burns exactly that copy at those times, and it is the path for silent clips. The video must be a public HTTPS URL.
curl -X POST https://api.sume.com/v1/video-captions \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: ai-label-001" \
-d '{
"video_url": "https://example.com/clip.mp4",
"style": "slam",
"cues": [{ "text": "AI-generated video", "start": 0, "end": 4 }]
}'Does a visible label satisfy anything?
Sume makes no claim that it does. Neither this license nor a platform's rules are settled by a caption; the docs describe how to draw text, not what text a rule needs. For the EU icon case, see that post.
Sources
Related posts
More in Use cases
- Export one video as 9:16, 1:1 and 16:9 with the API
Call Timeline 1.0 three times with the same clip and a different output.width and height each time. Three renders cost $0.30 for a clip under a minute.
- Add black frames to the start or end of a video with tpad
Use the tpad filter on Sume's video filter to add black frames before or after a clip. A few lines of graph, one encode job, a new MP4 with handles.
- Pinterest ad image size: 2:3, 1000x1500, 40-character title
Pinterest ad images: PNG or JPEG, 2:3 (1000x1500), 100-character title with about 40 showing. Why 1000x1500 clashes with Sume's x16 pixels rule.
- Pinterest Collections ads specs: one hero, up to 24 secondaries
A Pinterest Collections ad shows one hero and three secondaries, then up to 24 secondaries fullscreen. No GIFs, desktop only. A batching plan for the assets.
Written by Sume