RouteMux Docs
IntegrationsAgent Skills

Video

Make short videos from a text prompt with MiniMax H3 from your coding agent. The skill quotes the price and waits for your yes before it spends anything.

The routemux-video skill turns a text description into an MP4 with MiniMax H3. It always tells you the price first and waits for your go-ahead.

Not set up yet? Follow the quick start first. MiniMax H3 needs an account that has topped up at least once.

Ask for a video

Make a 5-second video of ocean waves at sunset and save it to ./waves.mp4

What happens next:

  1. It writes the prompt. Your request becomes a detailed prompt covering the action, the setting, the camera movement and the style.
  2. It quotes the price and waits. Nothing is submitted until you say yes.
  3. It generates and downloads. The job usually takes a few minutes. When it finishes, the MP4 is saved where you asked.

Options

Say what you want in plain words — "10 seconds", "vertical", "in 2K" — and the skill sets the values.

SettingChoicesDefault
LengthAny whole number of seconds from 4 to 155 seconds
Resolution768P or 2K (1080p is not offered)768P
Aspect ratio16:9, 9:16, 1:1, 4:3, 3:4, 21:916:9
You wantAsk for
A clip for YouTube or a website16:9 (the default)
TikTok, Reels or Shorts9:16
A square social post1:1
A cinematic letterbox look21:9
A final, best-quality version2K
A quick test of an idea4 seconds at 768P

Pricing

Video is billed per second, and 2K costs more per second than 768P:

cost = length in seconds × per-second price for the resolution

The current per-second prices are on the MiniMax H3 model page. You do not need to work it out yourself — the skill shows the exact quote before it submits anything.

If it takes a while

Some videos take longer than your agent lets a single command run. When that happens the skill prints a --resume <id> command and keeps waiting for the same job — it does not submit again, because a new submission would be billed again.

If your agent stops before the video arrives, ask it to keep waiting for the video rather than to try again.

Not available yet

The skill makes video from text only. Turning an image into a video is not available through the API yet.

On this page