MiniMax H3 API
MiniMax H3 by MiniMax for text to video, image to video, reference to video, 4–15s clips up to 2k with audio. Call it through Moyi API's OpenAI-compatible /v1/videos endpoint with one API key.
MiniMax ships H3 in its Hailuo AI video app; MiniMax describes it as the generation after Hailuo 01 and Hailuo 02. Here you call it by API. source
Loading price…
MiniMax H3 model IDs and parameters
| Mode | Model ID | Duration | Resolution | Aspect ratio | Audio |
|---|---|---|---|---|---|
| Text to video | minimax-h3-text-to-video | 4–15s | 768P, 2k | 16:9, 9:16, 1:1, 4:3, 3:4, 21:9 | Yes |
| Image to video | minimax-h3-image-to-video | 4–15s | 768P, 2k | 16:9, 9:16, 1:1, 4:3, 3:4, 21:9 | Yes |
| Reference to video | minimax-h3-reference-to-video | 4–15s | 768P, 2k | 16:9, 9:16, 1:1, 4:3, 3:4, 21:9 | Yes |
Call the MiniMax H3 API
1. Get an API key and set it as an environment variable
export TEAMO_API_KEY=sk-teamo-...2. Make your first request
Video generation is asynchronous: create a job, poll it, then download the MP4.
import os, time, requests
API = "https://api.moyi-api.com/v1/videos"
headers = {"Authorization": f"Bearer {os.environ['TEAMO_API_KEY']}"}
# 1. Create a video job
job = requests.post(API, headers=headers, json={
"model": "minimax-h3-text-to-video",
"prompt": "A small dog running on grass",
"seconds": "5",
}).json()
# 2. Poll until status is completed
while job["status"] not in ("completed", "failed"):
time.sleep(5)
job = requests.get(f"{API}/{job['id']}", headers=headers).json()
# 3. Download the MP4
if job["status"] == "completed":
video = requests.get(f"{API}/{job['id']}/content", headers=headers)
open("video.mp4", "wb").write(video.content)
else:
print(job.get("error"))For image to video, use minimax-h3-image-to-video and pass the first frame in input_reference.image_url (a URL or data URL).
Pricing
Loading price…
MiniMax H3 API FAQ
- What are the MiniMax H3 API model IDs?
- MiniMax H3 model IDs on Moyi API — Text to video: minimax-h3-text-to-video; Image to video: minimax-h3-image-to-video; Reference to video: minimax-h3-reference-to-video.
- Is MiniMax H3 the model in Hailuo AI?
- Yes. MiniMax's Hailuo AI site lists MiniMax H3 as live, and MiniMax's H3 post places it after Hailuo 01 and Hailuo 02. Here you call it through an API by model ID. source
- What is the difference between MiniMax H3 text-to-video and image-to-video?
- Text to video creates a clip from a prompt only. Image to video animates the image you pass in input_reference as the first frame, and reference to video uses your images as references for characters, objects or style.
- How long and what resolution can MiniMax H3 videos be?
- MiniMax H3 generates 4–15s clips up to 2k in 16:9, 9:16, 1:1, 4:3, 3:4, 21:9, with optional audio. Pass seconds as an integer string; if size is omitted, the model's first supported resolution and aspect ratio are used.
- How do I call the MiniMax H3 API?
- Send POST /v1/videos with the model ID and prompt, poll GET /v1/videos/{video_id} until status is completed, then download from /v1/videos/{video_id}/content. The endpoint follows OpenAI Videos, so the OpenAI SDK's client.videos.create works with base_url https://api.moyi-api.com/v1. Video generation API docs.
- How much does the MiniMax H3 API cost and how do I pay?
- MiniMax H3 is billed per second of generated video. New accounts start with a $0 balance — top up from $5 by card, then usage is deducted per job with no subscription. Per-second and per-clip prices are in the Pricing section of this page. Top up.