import requests, time
headers = {"Authorization": "Bearer " + AIMLAPI_KEY}
job = requests.post(
"https://api.aimlapi.com/v2/video/generations",
headers=headers,
json={
"model": "minimax/h3-max",
"prompt": "A serene timelapse of clouds over a mountain range"
},
).json()
gid = job["id"]
while True:
res = requests.get(f"https://api.aimlapi.com/v2/video/generations?generation_id={gid}", headers=headers).json()
if res.get("status") in ("completed", "error"):
break
time.sleep(5)
print(res["video"]["url"])
const headers = {
Authorization: `Bearer ${process.env.AIMLAPI_KEY}`,
"Content-Type": "application/json",
};
const job = await (await fetch("https://api.aimlapi.com/v2/video/generations", {
method: "POST",
headers,
body: JSON.stringify({
"model": "minimax/h3-max",
"prompt": "A serene timelapse of clouds over a mountain range"
}),
})).json();
let res;
do {
await new Promise((r) => setTimeout(r, 5000));
res = await (await fetch(`https://api.aimlapi.com/v2/video/generations?generation_id=${job.id}`, { headers })).json();
} while (["completed", "error"].includes(res.status) === false);
console.log(res.video.url);
# submit the generation — the response contains the job "id"
curl -X POST https://api.aimlapi.com/v2/video/generations \
-H "Authorization: Bearer $AIMLAPI_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"minimax/h3-max","prompt":"A serene timelapse of clouds over a mountain range"}'
# then poll until status is "completed" — video.url holds the result
curl "https://api.aimlapi.com/v2/video/generations?generation_id={id}" -H "Authorization: Bearer $AIMLAPI_KEY"
OpenAI-compatible — swap the base URL and it works with your existing SDK.
| Type | Price |
|---|---|
| Output | |
480p: $0.065/sec, 768p: $0.104/sec.
| Benchmark | Score | What it measures | Source | Retrieved |
|---|---|---|---|---|
| Text-to-Video Arena | 1117 | Human preference Elo from blind pairwise comparisons of generated videos, measured independently by Artificial Analysis | Source | October 1, 2026 |
MiniMax H3 Max is described in MiniMax API docs as a video generation model jointly released by MiniMax and fal.ai, post-trained by fal.ai on MiniMax H3 and optimized for high-speed generation.
It is presented as a separate video model from MiniMax H3, with support for mainstream 480P and 768P output and faster generation than MiniMax H3.
Joint release
MiniMax says MiniMax H3 Max was jointly released by MiniMax and fal.ai.
Post-trained variant
fal.ai describes H3 Max as a post-trained variant of MiniMax H3, tuned for stronger prompt adherence and better aesthetics.
Speed-focused optimization
MiniMax API docs say H3 Max is optimized for high-speed generation.
Supported outputs
MiniMax API docs say it delivers mainstream 480P and 768P output.
Current support
MiniMax API docs say H3 Max currently supports text-to-video and image-to-video, with reference generation coming soon.
Fast text-to-video creation
Use H3 Max for text prompts when fast video generation is the priority.
Image-to-video generation
Use H3 Max to turn images into video clips with the model’s current I2V support.
Short-form video iteration
H3 Max is positioned for quicker turnaround on mainstream-resolution video generation.
MiniMax H3 Max is documented for fast concept exploration and rapid creative iteration, especially when many short video alternatives need to be tested quickly.
It is also suited to social media videos and short-form clips, including TikTok, Reels, and Shorts, where speed and repeated prompt variations matter.
The model is also described as useful for ad concept testing and campaign variation work, making it a fit for marketers comparing multiple motion ideas before choosing a final direction.
Another documented use case is storyboard visualization and previsualization, where quick text-to-video or image-to-video output helps evaluate camera movement, scene flow, and visual direction early in production.
You can try this model in the browser before writing any code: open the AI/ML API playground, select the model and send a prompt. Nothing has to be installed locally — an API key is all you need. The same request is shown as cURL, Python and JavaScript in the code examples on this page, so a prompt that works in the playground can be copied straight into an application.
Open the playgroundLorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.Machine-readable API summary for LLM agents (llms.txt)
| Model | Input | Output | Context | Best for |
|---|---|---|---|---|
MiniMax H3 Max This page | Fast text-to-video and image-to-video generation | |||
| Native 2K narrative video with synchronized audio and consistent characters | ||||
| Video generation | ||||
| Video generation |
MiniMax H3 Max takes audio, image, text, video as input and returns video.
MiniMax H3 Max became available on August 26, 2026.
MiniMax H3 Max is priced at $0.065/sec.
MiniMax H3 Max is billed per generation — a fixed charge per output rather than by prompt length.
Yes, MiniMax H3 Max accepts image input as well as a text prompt.
MiniMax H3 Max was built by MiniMax.
Send a request to https://api.aimlapi.com/v2/video/generations with minimax/h3-max as the model id.
Yes. MiniMax H3 Max is served through AI/ML API, so the same key and endpoint format used for other models applies.
No. It outputs video at 480p or 768p resolution, not 2K.
Yes, it supports reference-to-video generation, letting you feed reference images or video to maintain consistency.
The shortest clip is 5 seconds; clips can run up to 15 seconds.
It generates video at 24 frames per second.
