Get started
Video model

MiniMax Video (Hailuo) — 2K clips with sound

Hailuo video models by MiniMax in GPTunneL: a clip of up to 15 seconds in 2K with sound in the same render. Pay per second, no subscription, no VPN.

What the Hailuo family is good at

The traits MiniMax keeps from generation to generation — you can count on them whichever model you pick.

Sound together with the picture

The top model produces stereo sound in the same render as the footage — speech, footsteps, room noise. No separate audio pass needed.

Motion physics

The reason the family gets picked for advertising: people and objects move with believable weight and inertia instead of drifting across the frame.

Video from text and from a frame

The model works both ways: it builds a scene from a description or brings a still to life — a photo, a concept, a product shot.

References as input

The top model takes more than text — images, video and audio set the style, the character or the voice so they stay consistent between takes.

Which model to pick

Three tiers in the family — for different tasks and budgets.

Hailuo 2.3 FastDrafts and idea huntingH3The default pickHailuo 2.3The proven generation
Best forSketching a scene, testing an idea, collecting optionsSocial clips and advertising insertsFamiliar workflows where the result is already dialled in
Resolution768p or 1080p2K768p or 1080p
Sound in the renderNoYesNo
Length6 or 10 secondsup to 15 seconds6 or 10 seconds
SpeedFastestMediumFast

Specs come from the platform catalogue. Exact per-model prices are on the Pricing

Every MiniMax video model

Release dates follow MiniMax's official announcements.

July 2026

H3

The current generation, also called Hailuo 3.0: a clip of up to 15 seconds in 2K with stereo sound in the same render. The model takes more than text as input — images, video and audio can set the style, the character or the voice.

October 2025

Hailuo 2.3Hailuo 2.3 Fast

The generation that took the family to 1080p: noticeably better motion physics, camera control and facial micro-expression. A faster Fast variant sits alongside it — the same generation, noticeably cheaper.

2025

Hailuo 02

The generation that put MiniMax on the map in video — on motion believability it caught up with far more expensive models.

How billing works

There is no subscription: you top up one balance and spend it on any model on the platform.

Pay per second

You are charged for the length of the finished clip and nothing else. Not using it costs nothing: no limits, no monthly fee.

One balance for every model

Hailuo, ChatGPT, Claude, image and music generation — all out of the same wallet. No separate subscription per service.

Top up the way you prefer

International cards, Apple Pay, Google Pay or crypto. The minimum top-up is $5.

See pricing

An example generation

The clip was generated in MiniMax from an ordinary text description. There are no tags or magic words to guess — the more precisely you describe the scene, the light and the camera move, the closer the result lands.

H3, generated from a text description

How to start with MiniMax Video

Sign in to GPTunneL
One account for every model on the platform.
A black bull emerges from the morning mist beside a sports car in a meadow, dew on the grass, the camera tracks slowly along the car
5 seconds2KH3

Sign in whichever way suits you.

The model switches right inside the input — you can change it between clips.

The finished video arrives at working resolution, with the sound in the same file.

Try MiniMax Video in GPTunneL

Signing up takes a minute, and $5 on the balance is enough to see whether the model fits your task.

Frequently asked questions

No. Requests go through GPTunneL's infrastructure, so the models open like any ordinary website — no VPN and no proxy.

Hailuo is the video family from the Chinese company MiniMax. It makes clips from a text description or from a still, and the latest generation also generates sound — speech, ambience and music in the same render as the footage.

The current H3 model (also called Hailuo 3.0) builds a clip of up to fifteen seconds in 2K with stereo sound, and takes more than text as input: attached images, video and audio set the style, the character or the voice. The previous Hailuo 2.3 generation works at 768p and 1080p and stays in the service as the cheaper option — it has no sound in the render.

The family's strong suit is motion physics: people and objects move with believable weight, which is why the clips end up in advertising inserts. The scene description is understood in dozens of languages, with no translation needed.

In GPTunneL the Hailuo models are available worldwide, without a VPN, with one balance and billing for the length of the finished clip. The same developer's text models are on the MiniMax page and share the same balance.