MiniMax H3 Flex API for Video Generation

Send model: minimax-h3-flex to POST /api/v1/videos/generations for text-to-video, image-to-video or first-and-last-frame video with native audio. It takes the same inputs and price as hailuo_h3, follows motion and action in the prompt more closely, and in our tests finishes each video in about 70% of the time hailuo_h3 takes.

Get API Key

Dancer caught mid-spin on a red paper backdrop under a hard studio light

Same Inputs as hailuo_h3, Closer Motion

Motion that follows the prompt

Describe the action step by step, such as a spin, a jump or a turn toward the camera, and the model plays it out between the first and last frame instead of only drifting the camera. The prompt is required and accepts up to 7,000 characters.

Read the MiniMax H3 Flex API docs

Gymnast mid-jump frozen between two chalk marks on a gym floor

First and last frame control

Pass image_url for the opening frame and optionally image_end_url for the closing frame. The output ratio follows the first frame; image_end_url must be sent together with image_url.

Read the MiniMax H3 Flex API docs

Two printed film frames pinned at either end of a storyboard wall

Audio-driven lip sync

Add audio_url and optionally audio_start: the characters sing or speak along with your audio, the soundtrack is your audio clip, and the video lasts exactly duration seconds. Works with text, a first frame or first + last frames, at no extra charge.

Read the MiniMax H3 Flex API docs

Singer in front of a vintage studio microphone with a glowing waveform behind her

Only model and prompt are required; everything else is optional.

MiniMax H3 Flex Request Fields and Limits

prompt (required)

Text describing the video, up to 7,000 characters.

image_url and image_end_url

Optional first-frame and last-frame images: JPG, PNG, WEBP or AVIF, up to 10 MB, shortest side at least 300 px. image_end_url must be used together with image_url.

resolution

480p, 720p or 2K; defaults to 2K. The resolution also sets the price, and 2K takes longer than 720p.

duration

Any integer from 5 to 15 seconds; defaults to 5.

ratio

Text-to-video only: 21:9, 16:9, 4:3, 1:1, 3:4 or 9:16, default 16:9. With image_url the output follows the input image.

audio_url, audio_start and callback

audio_url: MP3, WAV, M4A, AAC, FLAC or OGG up to 50 MB, the whole track is fine. audio_start: start offset in seconds. custom_callback_url: optional public HTTPS URL for status updates.

Billed per second of output, by resolution. The same rates as hailuo_h3.

MiniMax H3 Flex API Pricing

480p: 1.5 tokens ($0.03) per second

A 5-second 480p video costs 7.5 tokens ($0.15).

720p: 3 tokens ($0.06) per second

A 5-second 720p video costs 15 tokens ($0.30).

2K: 4.5 tokens ($0.09) per second

A 5-second 2K video costs 22.5 tokens ($0.45); 15 seconds at 2K costs 67.5 tokens ($1.35).

Lip sync at no extra charge, refunds on failure

Adding audio_url does not change the price. Tokens are deducted when the task is accepted and returned automatically if the generation fails; cost in the response is the amount actually charged.

Integrate the MiniMax H3 Flex API

1. Create an API key

Generate it at /keys and call the API from your server with Authorization: Bearer <key>.

2. Write the motion into the prompt

Describe what happens and in what order. For first-and-last-frame shots, describe how the scene gets from the first image to the last one.

3. Submit

POST {"model": "minimax-h3-flex", "prompt": "…", "image_url": "…", "resolution": "720p", "duration": 5} to /api/v1/videos/generations and store img_uuid.

4. Fetch the result

Poll GET /api/v1/jobs/detail or use custom_callback_url, and download the video before link_expired_at.

Create an API Key

Short clips where the action in the prompt has to actually happen.

What Developers Build with MiniMax H3 Flex

Transition and reveal effects

Give a first and last frame and describe the move between them, such as a spin, a jump through a backdrop or a costume change.

Music videos shot by shot

Generate each shot from its frames with the same song in audio_url and the shot's start time in audio_start, then join the shots in order.

Product and ad clips

Animate a product photo or a presenter shot into a 5 to 15 second clip with native sound.

Faster drafts at the same price

Swap hailuo_h3 for minimax-h3-flex in an existing pipeline to get results back sooner without changing request code or costs.

Differences from hailuo_h3, inputs and billing.

MiniMax H3 Flex API FAQ

How is it different from hailuo_h3?

Same fields, resolutions and price. It follows motion and action in the prompt more closely and, in our tests, finishes in about 70% of the time hailuo_h3 takes.

Can I use reference images?

No. MiniMax H3 Flex takes a first frame and an optional last frame. For reference-image driven video use hailuo_h3_refs.

Does the video have sound?

Yes, native audio is generated with the picture. With audio_url, the soundtrack is your audio clip instead.

What resolutions and lengths are available?

480p, 720p and 2K (default 2K), any whole number of seconds from 5 to 15.

How much does it cost?

1.5 tokens ($0.03), 3 tokens ($0.06) or 4.5 tokens ($0.09) per second for 480p, 720p and 2K.

Is GoEnhance affiliated with MiniMax?

No. GoEnhance Open API is an independent platform that provides unified access to video and image models. MiniMax is a trademark of its owner and is used only to identify the model.

Start Building with the MiniMax H3 Flex API

Create an API Key