Unlimited Open Source Models

Get Plan
Skip to main content
Imagen

MiniMax API — Hailuo Video & M2 LLM

Hailuo 2.3 and 02, MiniMax H3 video and the MiniMax M2 LLM family through the ModelsLab REST API — one key, no separate MiniMax account, per-call pricing shown before you send. Video plans from $21/month.

Get MiniMax API AccessAPI Documentation

What is the MiniMax API?

Two Product Lines, One Endpoint

MiniMax ships two things developers integrate: the Hailuo video models and the MiniMax M-series language models. The ModelsLab API carries both. Video goes through the same /api/v7/video-fusion/ endpoints as Kling, Veo, Wan and Seedance, and the LLMs go through the OpenAI-compatible chat endpoint — on one key, one bill and one integration.

Hailuo 2.3 is the current cost-effective generation at $0.067 per second of output, with a Fast image-to-video variant at $0.038 per second. Hailuo 02 and the MiniMax H3 family are priced per generation scaled by duration and resolution, from $0.096, and add start/end-frame and reference-to-video control for shots that need continuity.

What You Skip

No MiniMax account, no second set of credentials, no separate invoice, and no provider-specific SDK. Switching between MiniMax and any other video provider is a change to `model_id`; the request body, the auth header and the polling endpoint stay the same.

  • Hailuo 2.3, 2.3 Fast, 02 and 02 start/end-frame
  • MiniMax H3 text, image, reference and start/end-frame to video
  • MiniMax M1, M2, M2.1 and M2.5 on the LLM endpoint
  • Per-call price and its unit visible before you send
  • Async ids, fetch endpoint and webhook delivery
  • One API key shared with image, speech and LLM

MiniMax Models on ModelsLab

Pick the model with model_id on the video endpoint, or by name on the LLM endpoint.

model_idModelInputListed price and unit
Hailuo-2.3-t2vHailuo 2.3Text$0.067 / second
Hailuo-2.3-i2vHailuo 2.3Image$0.067 / second
Hailuo-2.3-Fast-i2vHailuo 2.3 FastImage$0.038 / second
Hailuo-02-t2vHailuo 02Textfrom $0.096, by duration
Hailuo-02-start-end-frameHailuo 02Start/end framesfrom $0.096, by duration
Minimax-h3-t2vMiniMax H3Textfrom $0.096, by duration
Minimax-h3-r2vMiniMax H3Reference imagesfrom $0.096, by duration
minimax-minimax-m2MiniMax M2 (LLM)Chatper million tokens

Prices as listed on each model page in September 2026. Note the unit: the Hailuo models are billed per second of output, the H3 models per generation scaled by duration and resolution. MiniMax is a third-party provider, billed per call from wallet balance on any plan.

MiniMax API quick start

Hailuo video from a prompt, from a still, and the M2 LLM on the same key.

Hailuo 2.3 text to video (Python)

Python
1import requests
2

3response = requests.post(
4    "https://modelslab.com/api/v7/video-fusion/text-to-video",
5    headers={
6        "Authorization": "Bearer YOUR_MODELSLAB_API_KEY",
7        "Content-Type": "application/json",
8    },
9    json={
10        "model_id": "Hailuo-2.3-t2v",
11        "prompt": "a cyclist crests a hill at sunrise, camera tracks alongside",
12        "duration": 5,
13        "aspect_ratio": "16:9",
14    },
15)
16

17# Returns an id — poll fetch/{id} or pass a webhook
18print(response.json())

MiniMax H3 image to video (JavaScript)

JavaScript
1const response = await fetch(
2  'https://modelslab.com/api/v7/video-fusion/image-to-video',
3  {
4    method: 'POST',
5    headers: {
6      'Content-Type': 'application/json',
7      Authorization: 'Bearer YOUR_MODELSLAB_API_KEY',
8    },
9    body: JSON.stringify({
10      model_id: 'minimax-h3-i2v',
11      prompt: 'slow dolly in, shallow depth of field',
12      init_image: ['https://example.com/still.jpg'],
13      duration: 5,
14    }),
15  },
16);
17

18const { id } = await response.json();
19console.log(`queued as ${id}`);

MiniMax M2 chat (Python, OpenAI SDK)

Python
1from openai import OpenAI
2

3client = OpenAI(
4    api_key="YOUR_MODELSLAB_API_KEY",
5    base_url="https://modelslab.com/api/v7/llm",
6)
7

8response = client.chat.completions.create(
9    model="minimax-minimax-m2",
10    messages=[{"role": "user", "content": "Draft three prompts for a product video."}],
11)
12

13print(response.choices[0].message.content)

Generate with MiniMax in 3 Steps

Same endpoint as every other video model.

STEP 01
STEP 01

Step 1: Get Your API Key

Create a ModelsLab account, subscribe to a video plan from $21/month, add wallet balance for third-party models, and copy your key.

STEP 02
STEP 02

Step 2: Send a MiniMax model_id

POST to /api/v7/video-fusion/text-to-video with Hailuo-2.3-t2v, Hailuo-02-t2v or Minimax-h3-t2v, plus prompt, duration and aspect_ratio. Use /image-to-video with init_image to animate a still.

STEP 03
STEP 03

Step 3: Fetch the Clip

The request returns an id. POST /api/v7/video-fusion/fetch/{id} until output is populated, or pass a webhook URL and receive the MP4 when it is ready.

Get API Key

Related video API guides

AI Video API

The full video endpoint set MiniMax sits inside: text, image, video and audio inputs.

Seedance API

ByteDance’s video family on the same endpoint, for a side-by-side comparison.

Kling API

Kling 3.0, V2.5 Turbo and motion control, priced per call on the same key.

Hailuo or H3?

They are two generations of the same provider’s video stack, and both are live. Hailuo 2.3 is the cheaper and faster option at $0.067 — or $0.038 for the Fast image-to-video variant — and is the sensible default for iteration and for volume. The H3 family costs $0.096 and adds reference-to-video and start/end-frame control, which is what you want when a shot has to match the one before it.

Because both sit behind the same endpoint, comparing them is a change to `model_id` and nothing else. Draft on 2.3 Fast, promote the take you are shipping to H3 if it needs the extra control.

The Self-Hosted H3 Builds Are Cheaper

Worth knowing before you pick a model_id: alongside the MiniMax-billed H3 rows, ModelsLab runs its own H3 builds on its own GPUs — h3-minimax-t2v and h3-minimax-r2v at $0.050 per second, and h3-minimax-ref-to-video-fast from $0.040. They are cheaper than the provider-billed equivalents and are covered by your plan rather than by wallet balance. If you do not specifically need MiniMax’s own hosting, start there.

Pricing

MiniMax is a third-party provider, so video generation is billed per call from wallet balance on top of a video plan, on every plan including the $199 unlimited tier — that tier covers the open-weight models ModelsLab self-hosts, and MiniMax is not one of them. Read the unit carefully: the Hailuo models are priced per second of output, so a five-second clip at $0.067 per second costs about $0.34, while the H3 models quote a base price that scales with duration and resolution. Every model page shows its own unit. LLM usage is billed per million tokens.

Why Call MiniMax Through ModelsLab

Key advantages that set us apart

Hailuo 2.3, 2.3 Fast, 02 and MiniMax H3 video
MiniMax M1, M2, M2.1 and M2.5 on the LLM endpoint
No MiniMax account or second invoice
Same endpoint as Kling, Veo, Wan and Seedance
Reference-to-video and start/end-frame control on H3
Per-call price and unit shown before you send
Async id, fetch endpoint and webhook delivery
Video plans from $21/month

Our Popular Use Cases

What teams build with the MiniMax API:

Hailuo 2.3 Fast at $0.038 per second makes short test cuts cheap enough to run dozens of variants.

High-Volume Social Clips

MiniMax API FAQ

Start on Hailuo 2.3. It is billed per second of output at $0.067, which makes iteration cheap, and for most prompts the quality difference is not what decides the shot. Move to H3 when you need reference-to-video or start/end-frame control to keep a subject consistent across clips.

Check the unit on the model page, not just the number. Hailuo is per second, so multiply by your duration — a five-second clip at $0.067 per second is about $0.34. The H3 models quote a base price that scales with duration and resolution. The final figure is shown before the request is sent.

Yes, and that is the main reason to call it here. Kling, Veo, Wan, Seedance and MiniMax all sit behind /api/v7/video-fusion/, so routing between them is a change to model_id with the same request body, the same key and the same fetch endpoint.

No. Unlimited applies to the open-weight video models ModelsLab self-hosts. MiniMax is third-party, so it is billed per call from wallet balance on every plan including that one.

No — the M-series LLMs are billed per million tokens on the OpenAI-compatible chat endpoint, while the video models are billed per call from wallet balance. Same key, two different meters.

Your Data is Secure: GDPR Compliant AI Services

ModelsLab GDPR Compliance Certification Badge

GDPR Compliant

Start Creating Amazing AI Videos in Minutes

Choose plan as per your needs, cancel anytime.

Coming Soon

We are making some changes to our pricing, please check back later.

Get Expert Support in Seconds

We're Here to Help.

Want to know more? You can email us anytime at support@modelslab.com

View Docs

The Hailuo video family (Hailuo 2.3, Hailuo 2.3 Fast and Hailuo 02, in text-to-video, image-to-video and start/end-frame variants), the MiniMax H3 family (text-to-video, image-to-video, reference-to-video and start/end-frame) and the MiniMax M1 and M2 LLMs.

MiniMax is a third-party provider, billed per call from wallet balance on top of a plan. In September 2026 the model pages list Hailuo 2.3 at $0.067 per second of output and Hailuo 2.3 Fast at $0.038 per second, while Hailuo 02 and MiniMax H3 start at $0.096 per generation scaled by duration and resolution. A five-second Hailuo 2.3 clip is therefore about $0.34, not $0.067. LLM usage is billed per million tokens.

No. ModelsLab calls MiniMax for you. You use a ModelsLab API key, the same one that covers Kling, Veo, Seedance, Wan and the rest of the video catalogue.

POST https://modelslab.com/api/v7/video-fusion/text-to-video with `model_id` set to `Hailuo-2.3-t2v`, `Hailuo-02-t2v` or `Minimax-h3-t2v`, plus `prompt`, `duration` and `aspect_ratio`. Use /image-to-video with `init_image` and the matching i2v model id to animate a still.

They are two generations of the same provider's video stack. Hailuo 2.3 is the cheaper and faster option at $0.067 per second of output, and the H3 family adds reference-to-video and start/end-frame control from $0.096 per generation scaled by duration. ModelsLab also self-hosts its own H3 builds — `h3-minimax-t2v` at $0.050 per second and `h3-minimax-ref-to-video-fast` from $0.040 — which are cheaper again because they run on our GPUs. All of them sit behind the same endpoint, so comparing them is a change to `model_id`.

Yes. MiniMax M1 and M2 are available on the OpenAI-compatible endpoint at /api/v7/llm/chat/completions, billed per million tokens, with the same API key that runs the video models.

ModelsLab is a paid service. Video plans start at $21/month and MiniMax is billed per call on top from wallet balance. Creating an account is free, but generating requires an active plan.

Explore Our Other Solutions

Unlock your creative potential and scale your business with ModelsLab's comprehensive suite of AI-powered solutions.

Imagen

AI Image Generation & Tools

Generate, edit, upscale, and transform images with state-of-the-art AI models.

Explore Imagen
Audio Gen

AI Audio Generation

Text-to-speech, voice cloning, music generation, and audio processing APIs.

Explore Audio Gen
Chat

Engage Seamlessly with LLM

Access powerful language models for chatbots, content generation, and AI assistants.

Explore Chat
3D Verse

Create Stunning 3D Models

Transform images and text into 3D models with advanced AI-powered generation.

Explore 3D Verse
Plugins

Explore Plugins for Pro

Our plugins are designed to work with the most popular content creation software.

Explore PluginsLearn More
API

Build Apps with
ML
API

Use our API to build apps, generate AI art, create videos, and produce audio with ease.

API DocumentationPlayground