Video

MiniMax H3 API

MiniMax H3 is MiniMax's general-purpose multimodal generation model, released on July 31, 2026, that generates video up to 15 seconds at 2K with native stereo audio. On APIMart you can call MiniMax-H3 for video, MiniMax-H3-Context-IR for prompt enhancement and MiniMax-H3-Regeneration to turn 768P previews into 2K.

2K VideoNative AudioMultimodal Reference

Get API Key API Docs

Service Highlights

Official Discounts
Pay-as-you-go
Fast & Low Latency

Input

Model

Generation Mode

No image: text-to-video. 1–2 images: first/last frame control. Mutually exclusive with Omni.

Prompt*

91 / 7000 characters

Duration5s

Resolution

Aspect Ratio

AIGC Watermark

First / Last Frame

First Frame

Last Frame

Content Moderation

Output

Video link valid for 72 hours

MiniMax H3 API Pricing

Transparent pricing with no hidden fees. Pay only for what you use.

Standard Generation

Variant	Tier	GoldCurrent price	Platinum	Diamond	Official	Your savings
`MiniMax-H3`	Reference image (first 5 free, then each)	0.2288Credits(~$0.0229)/pic	0.2174Credits(~$0.0217)/pic	0.2059Credits(~$0.0206)/pic	0.286Credits(~$0.0286)/pic	20%
	768P	0.5712Credits(~$0.0571)/sec	0.5426Credits(~$0.0543)/sec	0.5141Credits(~$0.0514)/sec	0.714Credits(~$0.0714)/sec	20%
	2K	0.9144Credits(~$0.0914)/sec	0.8687Credits(~$0.0869)/sec	0.823Credits(~$0.0823)/sec	1.143Credits(~$0.1143)/sec	20%

Reference Video

Added to generation duration · Reference videos support up to 15 seconds in total

Variant	Tier	GoldCurrent price	Platinum	Diamond	Official	Your savings
`MiniMax-H3`	768P · Reference video (per source second)	0.5712Credits(~$0.0571)/sec	0.5426Credits(~$0.0543)/sec	0.5141Credits(~$0.0514)/sec	0.714Credits(~$0.0714)/sec	20%
`MiniMax-H3`	2K · Reference video (per source second)	0.9144Credits(~$0.0914)/sec	0.8687Credits(~$0.0869)/sec	0.823Credits(~$0.0823)/sec	1.143Credits(~$0.1143)/sec	20%

Standard Generation

Variant	Tier	GoldCurrent price	Platinum	Diamond	Official	Your savings
`MiniMax-H3-Regeneration`	Reference image (first 5 free, then each)	0.1712Credits(~$0.0171)/pic	0.1626Credits(~$0.0163)/pic	0.1541Credits(~$0.0154)/pic	0.214Credits(~$0.0214)/pic	20%
`MiniMax-H3-Regeneration`	2K	0.3432Credits(~$0.0343)/sec	0.326Credits(~$0.0326)/sec	0.3089Credits(~$0.0309)/sec	0.429Credits(~$0.0429)/sec	20%

Reference Video

Added to generation duration · Reference videos support up to 15 seconds in total

Variant	GoldCurrent price	Platinum	Diamond	Official	Your savings
`MiniMax-H3-Regeneration`	0.3432Credits(~$0.0343)/sec	0.326Credits(~$0.0326)/sec	0.3089Credits(~$0.0309)/sec	0.429Credits(~$0.0429)/sec	20%

* Actual costs are subject to final output.

MiniMax-H3-Context-IR

Pricing Details

Transparent pricing with no hidden fees. Pay only for what you use.

Token Pricing

Per 1M tokens

Rate	GoldCurrent price	Platinum	Diamond	Official	Your savings
Input	6.6288~$0.6629	6.2974~$0.6297	5.9659~$0.5966	8.286~$0.8286	20%
Output	26.2865~$2.6287	24.9722~$2.4972	23.6579~$2.3658	32.8581~$3.2858	20%

MiniMax H3 API: 2K Video with Native Audio

Generate 4 to 15 second videos at 2K or 768P with MiniMax H3 on APIMart. Start from a prompt, a first and last frame, or up to 9 reference images, 3 videos and 3 audio clips. Enhance prompts with Context-IR and upgrade 768P previews to 2K with Regeneration.

Get API Key

50K+

Active Users

99.9%

Uptime

Faster

—

Cost Savings

Key Features of the MiniMax H3 API

What MiniMax H3 can do and what you control on APIMart.

Native 2K or 768P Output

2K is the default resolution. Choose 768P for cheaper previews and drafts.

4 to 15 Second Clips

Set duration to any whole number from 4 to 15 seconds. The default is 5.

Native Stereo Audio

MiniMax says H3 jointly models voice, sound effects and music, and all audio output is native stereo.

First, Last or First+Last Frame

Send first_frame_image, last_frame_image or both. In image-to-video mode, the aspect ratio follows your input image.

Multimodal References

Guide a clip with up to 9 reference images, 3 videos and 3 audio clips. Reference audio must be paired with an image or video, and references cannot be mixed with first/last frames.

Context-IR Prompt Enhancement

MiniMax-H3-Context-IR reads your idea and media and returns a structured, richer prompt as text. It does not create a video and is billed by tokens.

768P Preview, Then 2K Regeneration

MiniMax-H3-Regeneration turns a successful H3 768P task into 2K from its source_task_id. The 768P price plus the Regeneration price equals the direct 2K price.

Six Aspect Ratios, Long Prompts

Use 21:9, 16:9, 4:3, 1:1, 3:4 or 9:16. Text-to-video falls back to 16:9 when no ratio is set. Prompts can be up to 7,000 characters.

What You Can Build with the MiniMax H3 API

MiniMax positions H3 for advertising, e-commerce, product design and gaming content.

Ads with Dialogue and Sound

Create short ad spots where voice, music and sound effects come out with the picture.

E-commerce Product Videos

Use product photos as reference images to keep the product consistent across shots.

Storyboards and Transitions

Set a first and last frame to connect two keyframes in one shot.

Brand and Character Consistency

Reuse the same character or style images as references across a series of clips.

Game and App Trailers

Turn concept art and UI screens into short trailers and feature demos.

Preview-First Production

Iterate at 768P, then regenerate the approved take in 2K instead of re-rendering every draft at 2K.

How to Get a MiniMax H3 API Key

Three steps to your first MiniMax H3 video on APIMart.

Sign up and add funds

Create an APIMart account and top up your balance. The balance works for MiniMax H3 and every other model on the platform.

→

Create an API key

Generate an API key in the dashboard. The same key calls MiniMax-H3, Context-IR, Regeneration and MiniMax-H3-Max.

→

Send your first request

POST to /v1/videos/generations with model MiniMax-H3, or try the playground above first.

Start Creating

MiniMax H3 API Example: Text to Video and First/Last Frame Video

Call MiniMax H3 on APIMart with one POST request. Generation is asynchronous: submit a task, then poll the task endpoint to get the video URL.

1
Authenticate MiniMax H3 API requests
Save your APIMart key in the APIMART_API_KEY environment variable and send it as Authorization: Bearer <key> on every request.
2
Submit a MiniMax H3 task
POST /v1/videos/generations with model MiniMax-H3, your prompt, duration and resolution. Add first_frame_image and last_frame_image for image to video. The task ID is in data[0].task_id.
3
Get the MiniMax H3 video
Poll GET /v1/tasks/{task_id} until status is completed, then read the video URL from data.result.videos[0].url[0]. Results are kept for 72 hours, so download them in time.
See the MiniMax H3 API reference for every parameter, media limit and response field.

POST /v1/videos/generations

curl -X POST https://api.apimart.ai/v1/videos/generations \
  -H "Authorization: Bearer $APIMART_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "MiniMax-H3",
    "prompt": "A street musician plays violin in the rain at night, neon reflections on wet pavement, he smiles at the camera, cinematic lighting",
    "duration": 5,
    "resolution": "2K",
    "aspect_ratio": "16:9"
  }'

import os
import time
import requests

API_BASE = "https://api.apimart.ai"
headers = {"Authorization": f"Bearer {os.environ['APIMART_API_KEY']}"}

# 1. Submit the generation task
payload = {
  "model": "MiniMax-H3",
  "prompt": "A street musician plays violin in the rain at night, neon reflections on wet pavement, he smiles at the camera, cinematic lighting",
  "duration": 5,
  "resolution": "2K",
  "aspect_ratio": "16:9"
}
resp = requests.post(f"{API_BASE}/v1/videos/generations", headers=headers, json=payload, timeout=60)
resp.raise_for_status()
task_id = resp.json()["data"][0]["task_id"]

# 2. Poll the task until it finishes
for _ in range(200):
    task_resp = requests.get(f"{API_BASE}/v1/tasks/{task_id}", headers=headers, timeout=30)
    task_resp.raise_for_status()
    task = task_resp.json()["data"]
    if task["status"] == "completed":
        print(task["result"]["videos"][0]["url"][0])
        break
    if task["status"] == "failed":
        raise RuntimeError(task.get("error", {}).get("message", "generation failed"))
    time.sleep(10)
else:
    raise TimeoutError(f"task {task_id} did not finish in time; continue GET polling with this task ID")

curl -X POST https://api.apimart.ai/v1/videos/generations \
  -H "Authorization: Bearer $APIMART_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "MiniMax-H3",
    "prompt": "The camera slowly pushes in as the scene moves from the first frame to the last frame, natural motion and ambient sound",
    "first_frame_image": "https://example.com/start.jpg",
    "last_frame_image": "https://example.com/end.jpg",
    "duration": 5,
    "resolution": "768P"
  }'

TASK_ID="task_xxx"  # replace with the task_id returned by the POST request
curl "https://api.apimart.ai/v1/tasks/$TASK_ID" \
  -H "Authorization: Bearer $APIMART_API_KEY"

MiniMax H3 API Parameters

Parameter	Type	Required	Default	Description
`model`	string	Required	Default: MiniMax-H3	Use MiniMax-H3.
`prompt`	string	Required	Default: —	Scene, subject, motion and style, up to 7,000 characters. Required in every mode.
`duration`	integer	Optional	Default: 5	Whole number from 4 to 15 seconds.
`resolution`	string	Optional	Default: 2K	2K or 768P.
`aspect_ratio`	string	Optional	Default: —	21:9, 16:9, 4:3, 1:1, 3:4 or 9:16. Text-to-video falls back to 16:9; image-to-video follows the input image.
`first_frame_image`	string	Optional	Default: —	First frame image URL. Cannot be combined with reference fields.
`last_frame_image`	string	Optional	Default: —	Last frame image URL. Can be used with first_frame_image.
`image_urls`	string[]	Optional	Default: —	Up to 9 reference images.
`video_urls`	string[]	Optional	Default: —	Up to 3 reference videos, 2–15 seconds each, 15 seconds in total.
`audio_urls`	string[]	Optional	Default: —	Up to 3 reference audio clips. Must be paired with a reference image or video.
`watermark`	boolean	Optional	Default: false	Add an AIGC watermark.
`nsfw_check`	boolean	Optional	Default: false	Review the prompt and input images with omni-moderation-latest before submission.

User Reviews for MiniMax H3

Real feedback from creators and developers using MiniMax H3 worldwide. These reviews describe the original MiniMax H3 series experience.

“The consistency of MiniMax H3 is amazing! Character images remain stable across multiple clips, and the 2K quality looks production-ready.”

Wei Zhang

Independent Animator

“As a developer, I value stability and speed. MiniMax H3 on APIMart delivers great performance with multimodal controls that fit our pipeline.”

David Chen

Full-Stack Engineer

MiniMax H3 API FAQ

Answers about MiniMax H3 parameters, modes, pricing and integration on APIMart.

What is MiniMax H3?

MiniMax H3 is a general-purpose multimodal generation model that MiniMax released on July 31, 2026. It uses a new architecture rather than the Hailuo 02 design and generates video up to 15 seconds at 2K with native stereo audio. MiniMax says early testing shows it is ready for commercial content creation such as ads, e-commerce and gaming. On APIMart you call it with the model ID MiniMax-H3.

Is MiniMax H3 the same as Hailuo 3 or Hailuo 03?

MiniMax's official name for the model is MiniMax H3 (model ID MiniMax-H3), released on July 31, 2026 as the next generation after Hailuo 01 and Hailuo 02. MiniMax offers it on Hailuo AI (hailuoai.video). MiniMax's announcement does not use the names Hailuo 3 or Hailuo 03, but many sites and users call MiniMax H3 by those names. On APIMart, call it with MiniMax-H3 or MiniMax-H3-Max.

What parameters does the MiniMax H3 API support?

model (MiniMax-H3) and prompt (up to 7,000 characters) are required. Optional: duration (4–15, default 5), resolution (2K or 768P, default 2K), aspect_ratio (21:9, 16:9, 4:3, 1:1, 3:4 or 9:16), first_frame_image, last_frame_image, image_urls, video_urls, audio_urls, image_with_roles, watermark (default false), nsfw_check (default false) and webhook.

Which generation modes does MiniMax H3 support?

MiniMax H3 picks the mode from your fields, so there is no mode parameter. A prompt alone runs text-to-video. first_frame_image and last_frame_image run image-to-video. image_urls, video_urls and audio_urls run multimodal reference-to-video. Frame fields and reference fields cannot be combined, and audio cannot be used alone.

What is MiniMax H3 Context-IR?

Context-IR (model MiniMax-H3-Context-IR) is a prompt enhancement step. You send your idea plus optional images, video or audio, and it returns a structured, richer prompt in result.prompt, with no video. Pass that prompt as-is to MiniMax-H3. APIMart docs say it typically completes in 20–40 seconds, and it is billed by tokens.

How does MiniMax H3 Regeneration to 2K work?

Generate a 768P preview with MiniMax-H3, then call MiniMax-H3-Regeneration with that task's source_task_id. The platform fills in the prompt, media and source video and outputs 2K. It is not a general upscaler: the source must be a successful MiniMax-H3 768P task from your account. MiniMax-H3-Max output cannot be used as a source.

How much does the MiniMax H3 API cost?

As of October 2026, Gold member prices are $0.05712 per second at 768P and $0.09144 at 2K, so a 5-second 2K clip costs about $0.46. The first 5 input images are free, then $0.02288 per image; reference video seconds are billed at the output rate. Regeneration costs $0.03432 per second, and Context-IR $0.66 per million input tokens and $2.63 per million output tokens. See the pricing table on this page for current rates.

What is the difference between MiniMax H3 and MiniMax H3 Max?

MiniMax-H3 outputs 2K or 768P, runs 4–15 seconds and works with Context-IR and 2K Regeneration. MiniMax-H3-Max outputs 480P, 768P or 1080P, runs 5–15 seconds and does not support 2K or Regeneration. Both support text-to-video, first/last frames and multimodal references. See the MiniMax H3 Max API page for details.

What is the difference between MiniMax H3 and Hailuo 2.3?

MiniMax H3 is the newer generation. It adds 2K output, clips up to 15 seconds, native audio, last-frame control and multimodal references with images, video and audio. Hailuo 2.3 outputs 768p or 1080p clips of 6 or 10 seconds from text or a single first frame, at a lower price per second.

Is MiniMax H3 open source?

At launch, MiniMax said it plans to open the H3 model weights, subject to applicable laws and regulations. Check MiniMax's official channels for the current release status and license. APIMart provides hosted API access to MiniMax H3, so you do not need to run the model yourself.

How do I call the MiniMax H3 API?

POST /v1/videos/generations with your APIMart key, model MiniMax-H3 and a prompt. The response returns a task ID in data[0].task_id. Poll GET /v1/tasks/{task_id} until the status is completed, then read the video URL. See the code example on this page.

Is the MiniMax H3 API free?

No. There is no free quota for the MiniMax H3 API. It is pay-as-you-go, you only pay for successful requests, and invalid parameters are rejected without charge. Gold members get 20% off, and Platinum and Diamond members save up to 28%. Create a key on the API Keys page of your dashboard to start.

MiniMax H3 API

MiniMax H3 API Pricing

Token Pricing

MiniMax H3 API: 2K Video with Native Audio

Key Features of the MiniMax H3 API

What You Can Build with the MiniMax H3 API

How to Get a MiniMax H3 API Key

Sign up and add funds

Create an API key

Send your first request

MiniMax H3 API Example: Text to Video and First/Last Frame Video

Authenticate MiniMax H3 API requests

Submit a MiniMax H3 task

Get the MiniMax H3 video

MiniMax H3 API Parameters

User Reviews for MiniMax H3

MiniMax H3 API FAQ

What is MiniMax H3?

Is MiniMax H3 the same as Hailuo 3 or Hailuo 03?

What parameters does the MiniMax H3 API support?

Which generation modes does MiniMax H3 support?

What is MiniMax H3 Context-IR?

How does MiniMax H3 Regeneration to 2K work?

How much does the MiniMax H3 API cost?

What is the difference between MiniMax H3 and MiniMax H3 Max?

What is the difference between MiniMax H3 and Hailuo 2.3?

Is MiniMax H3 open source?

How do I call the MiniMax H3 API?

Is the MiniMax H3 API free?

Related Models

Latest MiniMax H3 articles

Key Features of the MiniMax H3 API

What You Can Build with the MiniMax H3 API

How to Get a MiniMax H3 API Key

Sign up and add funds

Create an API key

Send your first request

User Reviews for MiniMax H3

MiniMax H3 API FAQ

What is MiniMax H3?

Is MiniMax H3 the same as Hailuo 3 or Hailuo 03?

What parameters does the MiniMax H3 API support?

Which generation modes does MiniMax H3 support?

What is MiniMax H3 Context-IR?

How does MiniMax H3 Regeneration to 2K work?

How much does the MiniMax H3 API cost?

What is the difference between MiniMax H3 and MiniMax H3 Max?

What is the difference between MiniMax H3 and Hailuo 2.3?

Is MiniMax H3 open source?

How do I call the MiniMax H3 API?

Is the MiniMax H3 API free?

Related Models