MiniMax H3 Fast

New model · faster HaiLuo. Same settings as the main line.

H3 speed tier. 4–15s, 768p / 2K.

Just launched

Sixteen video lines

Grok, Seedance, MiniMax, Veo, Kling, Wan, and HappyHorse are all here. Open a card for that model's page.

Kling V4

Kling 4, 30s per take

Kling V4. 3–30s, 720p / 1080p. Up to 10 references. Drops aspect ratio when images are present.

Need longer and more refs? Switch to V4.

Kling 43–30s10 refs

Specs

Speed tier, same cap

Same 2K as the main line. Need 1080p? Switch to Normal.

Model ID
minimax-h3-fast
Duration
4–15 seconds
Resolution
768p / 2K
References
Up to 2
01

Faster

Same H3 line, speed first.

02

Same resolution

768p / 2K.

03

Switch

Need steadier? Normal. Need longer? Max.

Into your workflow

Use MiniMax H3 Fast now

Generate in the dashboard, or take the same Model ID into the API.

1Web

Open AI video

Pick this model in the dashboard, write a prompt, and generate.

2Connect

Set endpoint and model

OpenAI-compatible API. Paste the exact Model ID.

https://api.aicodewith.ai/v1
3Make it real

Send the first request

Model ID: minimax-h3-fast

minimax-h3-fast
Quickstart
Need the full API notes?
Open the video API page for the parameter table and sample calls.
Open API page