What Hailuo AI is
Hailuo AI is a video generation service from MiniMax, an AI company based in Shanghai. MiniMax was founded in December 2021 and listed on the Hong Kong Stock Exchange on 9 January 2026. You describe a scene, or upload a photo, and Hailuo draws a short clip. Wikipedia says MiniMax launched Hailuo AI in March 2024, its first video model, video-01, in September 2024, and audio functions on 20 January 2025.
Hailuo runs in a web browser at hailuoai.video, and MiniMax also lists a desktop app. The newest model is MiniMax H3. MiniMax's model card calls it a general-purpose, omni-modal system that takes text, images, video and audio and makes video with native stereo audio, from 4 to 15 seconds, at up to 2K. Hailuo 2.3 is the model that most older tutorials show.
Use is freemium. The video generator lesson reads Hailuo's subscription terms as giving welcome credits that expire 3 days after you get them, with paid plans above that.
What Hailuo AI is good for
- Clips from a text prompt: a baker at dawn, a product turning on a table. Tutorials call this Text to Video.
- A photo that moves: upload a start picture and describe only the motion. Tutorials call this Image to Video.
- A set start and end: H3 takes zero, one or two images. One image is the first or last frame, and two images are both, which Hailuo's tutorials call Start to End Frame.
- Sound and speech in the same clip: H3 makes native stereo audio, and its card lists stable dialogue in 11 languages, among them English and Chinese.
- The same character across shots: tutorials show Subject References, and one creator's workflow builds consistent character pictures before the scenes.
- Many inputs at once: the omni-reference mode takes up to 9 images, 3 video clips and 3 audio clips, with 12 files at most.
When to use Hailuo AI, and when to pick another tool
| Your job | Use Hailuo? | Or try |
|---|---|---|
| A short clip with sound and speech | Yes. H3 makes native stereo audio and dialogue in 11 languages | ElevenLabs, if you only need the voice |
| A free first try at AI video | Possible, but welcome credits expire after 3 days and free downloads carry a watermark | Runway or Adobe Firefly, which list free allowances |
| A clip from your own photo with a set start and end | Yes. H3 takes one or two images as first and last frames | Kling |
| A clip of a famous character or a brand logo | No. Three studios sued over Hailuo AI in September 2025, and a 2024 report says it reproduced TV channel logos | Your own characters and brands |
| Video that runs on your own computer | Possible with the H3 weights, on strong GPUs and inside the license limits | Wan2.2 in ComfyUI |
| A still picture with words in it | Not the best fit | Ideogram |
How to use Hailuo AI, step by step
These steps are for Hailuo's website. Menus move when the app updates, so they use the feature names its tutorials use and leave the rest for you to follow on screen.
- Open hailuoai.video, the address MiniMax lists on its H3 model card, then sign up or log in. Follow the sign-in options on screen.
- Check your credits before you start. The video generator lesson says welcome credits expire 3 days after you get them, so plan your first clips for those days.
- Open the video tool and choose how to start. Franklin AI's chapters name the two ways Text to Video and Image to Video, so look for those names or ones close to them.
- For Image to Video, upload a start picture. A clear photo or illustration with one main subject gives the clip something firm to move. Add a second picture only if you want a set last frame.
- Choose the model. Pick MiniMax H3 if the menu lists it, or the newest model it names.
- Type your prompt: the style, one shot, the camera move and the sound. The example below shows the order.
- Set the length and the shape. The H3 card lists 4 to 15 seconds, and shapes from 21:9 and 16:9 to 1:1 and 9:16. Check the credit cost near the button.
- Press the generate button and wait for the clip to render.
- Watch the whole clip with the sound on. Change one part of the prompt and generate again, then download the clip you like. Free downloads carry a watermark.
Open the video tool and choose how to start: Text to Video from words, or Image to Video from a picture.
Open the model menu and pick MiniMax H3 if it is listed, or the newest model it names.
Type your prompt in the box: the style, one shot, the camera move and the sound.
Set the length and the shape, check the credit cost on the generate button, then press it and wait for the clip.
AI-made illustrations of Hailuo AI's screens, drawn on 6 November 2026 to match these steps. They are not screenshots, so the real app may look different: follow what you see.
The tutorials at the top of this page show these steps on screen. Hailuo changes its menus now and then, so if a name differs, follow the one you see.
An example to try
Paste this prompt after you choose Text to Video. It follows the order of MiniMax's own prompt guide for H3: style, a shot, a camera move with its size and speed, then the sound.
Live-action, cinematic. A medium wide shot of a small bakery at dawn. The camera pushes in with small amplitude at slow speed toward the front counter, where a baker slides a tray of bread out of the oven. Warm light, steam rising. Sound: the scrape of the tray, a low oven hum, quiet street noise outside. No music, no text on screen.- One medium wide shot of a small bakery at dawn, in a live-action cinematic style.
- The camera pushes in slowly toward the counter, and a baker slides a tray of bread out of the oven.
- Warm light and rising steam fill the picture.
- The sound is the scrape of the tray, an oven hum and quiet street noise. No music and no text appear.
- If the camera does not move the way you wanted, change only the camera sentence and generate again.
Tips for better results
- Open with the style. MiniMax's prompt guide lists Cinematic, live-action, 2D-animated, 3D CG, claymation, watercolor and vintage film.
- Write the camera move as a type, a size and a speed, such as "pushes in with small amplitude at slow speed". The guide lists push in, pull out, pan, tilt, arc shot, tracking shot and static shot.
- Describe sound apart from the picture: ambient sounds, the sounds of actions, and any music the characters cannot hear.
- For speech, write the words in their own language and say who speaks. H3's card lists Arabic, Chinese, English, French, German, Italian, Japanese, Korean, Portuguese, Russian and Spanish as stable.
- For a cut inside one clip, the guide opens each later shot with a time, such as "At 00:03.500, the camera cuts to". Keep a clip to one shot until that works.
- Start from a picture when the look matters. The Rourke tutorial uses Image to Video to lock in style and texture.
- Try the shortest length first, since the H3 card lists 4 seconds as the minimum. A bad prompt then costs fewer of your credits.
What to know before you pay
- Credits expire. The video generator lesson reads Hailuo's terms as welcome credits that last 3 days and plan credits that last a month. The terms it read were last updated in July 2025, so check the plan page.
- Free downloads carry a watermark. On a paid plan, the lesson reads the terms as Hailuo claiming no ownership and allowing commercial use. Read them before a clip goes into an ad.
- Hailuo moderates what you send. The H3 card says text, images, videos and enhanced prompts go through automated checks, and content suspected of being unlawful, pornographic or infringing may be blocked.
- Copyright is a live issue. Wikipedia reports that Disney, Universal and Warner Bros. Discovery sued in September 2025, alleging Hailuo AI infringed their characters, and that Broadcast reported in December 2024 that it could reproduce British TV channel logos.
- The open weights have limits by region. MiniMax's license Q&A says the license is restricted in the EU, UK, South Korea and US for now, where organizations can apply for a formal license, while the API is open worldwide.
- Names move. MiniMax's own YouTube channel for Hailuo is called MiniMax Design, and models change often, so follow the model names shown in the app.
- A photo of a real person needs that person's consent.
Learn more about Hailuo AI
- 2Best AI video generator in 2026: which one to use for each jobWhich AI video generator to use: Google Veo and Gemini Omni, Runway, Kling, Luma, Pika, Hailuo, Seedance and Wan compared on free plans, prices and limits.
- 3How to make AI videos with Gemini and Google FlowHow to make AI videos with Gemini Omni and Google Flow: describe one shot, set the camera and the sound, start from your own photo and know the clip lengths.
Frequently asked questions
Is Hailuo AI free?
Partly. The video generator lesson reads Hailuo's subscription terms as giving welcome credits that expire 3 days after you get them, and free downloads carry a watermark. Paid plans sit above that, so check the plan page for the current allowance.
Who makes Hailuo AI?
MiniMax, an AI company in Shanghai founded in December 2021. It listed on the Hong Kong Stock Exchange on 9 January 2026, and Wikipedia names Hailuo AI as its video generation service.
What is MiniMax H3?
MiniMax's newest video model, which Hailuo's web app offers. Its card calls it an omni-modal system that takes text, images, video and audio and makes video with native stereo audio, from 4 to 15 seconds, at up to 2K.
How long can a Hailuo video be?
It depends on the model. The H3 card lists 4 to 15 seconds, and the video generator lesson's chart lists 10 seconds for Hailuo 2.3 through the API.
Can I use Hailuo videos in ads?
On a paid plan, the lesson reads Hailuo's terms as allowing commercial use, and free downloads carry a watermark. Read the terms yourself, and keep famous characters and logos out of the prompt.
Can I run Hailuo on my own computer?
Partly. The H3 weights are public, and MiniMax's announcement says they were verified on GPUs including the RTX 5090 and RTX 6000. The card's own server example uses 4 GPUs, and the license has limits by region.
Does Hailuo make videos with sound?
Yes, with H3. Its card lists native stereo audio at 32 kHz and stable dialogue in 11 languages. Describe the sound in your prompt apart from the picture.
Sources
- Wikipedia: MiniMax Group https://en.wikipedia.org/wiki/MiniMax_Group
- Hugging Face: MiniMaxAI/MiniMax-H3 https://huggingface.co/MiniMaxAI/MiniMax-H3
- MiniMax H3: Video prompt writing guide https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/docs/VIDEO_PROMPT_WRITING_GUIDE_base_en.md
- MiniMax H3: Q&A about the license https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/docs/QA-about-License.md
- Hailuo AI: Video subscription service terms https://hailuoai.video/doc/payment-policy.html
- MiniMax API docs: Video pricing https://platform.minimax.io/docs/guides/pricing-video
- Runway: Pricing https://runway.com/pricing
- Adobe: Firefly plans https://www.adobe.com/products/firefly/plans.html
- GitHub: Wan-Video/Wan2.2 https://github.com/Wan-Video/Wan2.2
- MiniMax Design on YouTube: MiniMax-H3 Is Now Publicly Available https://www.youtube.com/watch?v=AvjamLzt2pg
- Franklin AI on YouTube: The Ultimate MiniMax Hailuo Guide https://www.youtube.com/watch?v=aI82rYGiRk0
- Artificial Quotient on YouTube: MiniMax H3 Tutorial https://www.youtube.com/watch?v=-HcMguqI4ts
- Rourke on YouTube: Hailuo AI Video Masterclass https://www.youtube.com/watch?v=BnwcL1mhoPY