Hailuo 3 · online studio
This MiniMax H3 AI video generator turns Hailuo 3 text or a still into an MP4. Buy credits to generate. A purchase unlocks 768p, and credits unlock 10–15s. Independent tool — not official MiniMax or Hailuo.
Generating your video…
Warming up the model…
Something went wrong.
Motion demo — not MiniMax H3 output
Google sign-in · Credit packs · Download as MP4
The model
MiniMax H3 — also called Hailuo 3 or Hailuo 3.0 — is MiniMax’s multimodal video model. This page is an independent MiniMax H3 AI video generator: type a prompt or upload a still, get a short MP4.
MiniMax shipped the model in 2026 as a general-purpose stack: one system that can read text, images, video, and audio together and return a short clip with picture and sound. Official write-ups list about 4–15 seconds at 24 fps, native 32 kHz stereo, a 768-pixel short-edge default, and a separate 2K regenerate pass (H3-Regenerate-2K) that re-runs the 768p take in context instead of stretching pixels. That is the family people mean when they search Hailuo 3.
Hailuo is the product brand; H3 is the third generation after Hailuo 2.3. 2.3 was known for image-to-video around 768p or 1080p and clips closer to 6–10 seconds, usually without native stereo. Hailuo 3 is sold as native 2K, longer takes, omni references (up to nine images, three videos, three audio clips), first-and-last frame control, and stronger instruction following — camera language, on-screen text, brand boards. Search Hailuo 3, Hailuo 3.0, Hailuo H3, and MiniMax Hailuo 3 and you are looking at the same line.
This MiniMax H3 AI video generator is a smaller hosted studio for that search. You stay on minimaxh3.lol: text to video, or stills as references, 5 / 10 / 15 seconds, after you buy a pack. A purchase unlocks 768p. We do not run official 2K, first-and-last-frame pairs, nine-image omni, or a separate lip-sync upload here. If you need those, use MiniMax’s own Hailuo app or another host that actually exposes them.
Official MiniMax materials describe MiniMax H3 as a general-purpose stack: text, images, video, and audio in one context, then a clip with picture and sound out. Older Hailuo 2.3 workflows were closer to “one still in, silent motion out.”
GitHub and MiniMax docs set the MiniMax H3 default short edge at 768 pixels. 2K is a second pass (H3-Regenerate-2K) that re-runs the take with the original context — not a simple upscale. This MiniMax H3 AI video generator sells that 768p pass after a purchase.
MiniMax H3 is built to return 32 kHz stereo with the frames: room tone, effects, sometimes dialogue. This page downloads the MP4 as the stack returns it. We do not take a separate voice file for lip-sync.
Official MiniMax H3 / Hailuo 3 can take up to nine images, three video clips, and three audio clips, or a first and last frame. Those modes are real on MiniMax’s app and on some API hosts. They are not switches on minimaxh3.lol.
Write the subject, setting, one action, and camera move. This MiniMax H3 AI video generator renders a short MP4 — no timeline or stock footage required. Pick 16:9, 9:16, or 1:1 before you run.
Drop in a PNG, JPG, or WebP (up to 8 MB). The still becomes the opening frame; your prompt describes how it should move. Aspect follows the photo. A last-frame landing shot is not a control here.
There is no guest trial and no signup gift. Buy a pack, then generate. A purchase unlocks 768p. Official 2K marketing is real for MiniMax; it is not a switch on this page.
Honest limits
Hailuo 3 is a wide model. This browser tool sells the text-or-image path we actually call: 480p until you pay, then 768p.
| Capability | Official Hailuo 3 family | This MiniMax H3 AI video generator |
|---|---|---|
| Text to video | Yes — prompt-only clip | Yes — this is the default mode |
| Image to video | First frame, last frame, or both | One reference still as the opening frame |
| Omni references | Up to 9 images, 3 videos, 3 audio clips (12 files max) | Not on this page |
| Duration | About 4–15 seconds at 24 fps | 5, 10, or 15 seconds |
| Resolution | 768p default; 2K via in-context regenerate (H3-Regenerate-2K) | 768p after a purchase — we do not sell 2K here |
| Audio | Native 32 kHz stereo with the picture | MP4 as returned; no lip-sync or voice upload |
| Aspect ratios | 21:9 through 9:16, including 1:1 and 4:3 | 16:9, 9:16, and 1:1 on text to video |
| Instruction edits | Change one element and keep the rest of the take | No separate edit mode — generate a new clip |
| Open weights | H3-Base on GitHub / Hugging Face; 2K regenerate often API-only | Hosted generator only — we do not ship weights |
Naming
If you used Hailuo 2.3, MiniMax H3 (Hailuo 3) is the jump: longer takes, native stereo, claimed 2K, and omni references. This generator still sits on the text-or-image path we can actually call: 480p until you pay, then 768p.
Usually remembered as image-to-video around 768p or 1080p, clips nearer 6–10 seconds, and no native stereo in the same pass. Fine for a moving still; weaker than MiniMax H3 as a “write the whole shot” studio.
MiniMax H3 is sold as 4–15s at 24 fps, native stereo, claimed 2K, omni references, first-and-last-frame control, and stronger instruction following (camera language, on-screen text, brand boards).
This MiniMax H3 AI video generator runs the text-or-one-still path we can actually call: 768p after a purchase, 5 / 10 / 15 seconds, 16:9 · 9:16 · 1:1 on text to video. No fake 2K toggle.
Four steps
Four steps from idea to a MiniMax H3 MP4. The same flow works for text to video and image to video.
Start from text only, or drop in a still. Image to video uses that photo as the opening frame.
Write subject, setting, action, style, and camera move. Pick 5s, 10s, or 15s before you generate.
This MiniMax H3 AI video generator renders a short MP4. Buy a pack first. A purchase unlocks 768p.
Watch the clip on the page, then download the MP4. Signed-in videos also land in My works.
In the browser
What this MiniMax H3 AI video generator actually does in the browser — not a list of official MiniMax lab specs.
Write a scene and this MiniMax H3 AI video generator returns a short MP4. No timeline, stock footage, or desktop app.
Upload a PNG, JPG, or WebP (up to 8 MB). MiniMax H3 image to video uses that still as the first frame; your prompt describes how it should move.
Sign in with Google, buy a pack, then generate. There is no guest trial and no signup gift. Failed runs are not charged.
A purchase unlocks MiniMax H3 at 768p and Turbo at 720p. Credits are 5 / 8 / 10 for 5s / 10s / 15s on H3.
Work here is a prompt, a duration, and Generate. Failed runs are not charged when Account reports a failure.
Nothing to install. Generate on desktop or mobile and download the file as a plain MP4.
Studio demos
Atmosphere loops for the page, not MiniMax H3 output. Generate your own MP4 in the tool above — text to video or image to video.
One MiniMax H3 AI video generator, several short-clip jobs. Each card is a setting recipe, not a slogan.
Scenario: a Reels / Shorts / TikTok draft. Input: one MiniMax H3 text to video line. Setting: 9:16, 5–15s after you buy. Output: a 768p vertical beat. Caveat: one action — not a whole storyboard.
Scenario: a landing hero or paid-social test. Input: PNG/JPG pack shot plus a MiniMax H3 motion line. Setting: image to video, 5–10s. Output: the still as the opening frame. Caveat: aspect follows the photo; we do not take a last-frame product shot.
Scenario: look-dev before a crew. Input: MiniMax H3 lighting, lens, and camera language. Setting: 16:9, 10–15s. Output: a moving sketch. Caveat: paid 768p is a board, not a 2K finish.
Scenario: one subject, one beat. Input: who they are and the single action. Setting: 5–10s, locked-off or slow push. Output: a readable close-up. Caveat: omni character sheets (nine refs) are not on this page.
Scenario: illustrate a process without hunting stock. Input: the step you need to show. Setting: 16:9, 5–10s. Output: a short plate you can talk over. Caveat: on-screen text is not guaranteed; keep titles for the editor.
Scenario: a moving plate under a track. Input: atmosphere and camera. Setting: 5s loop-friendly or 15s paid. Output: MP4 to drop on a timeline. Caveat: sync and cuts are your job; this tool only makes the plate.
Prompt-first MiniMax H3 clips, and credits that stay on this site.
Buy a pack or subscribe, then generate MiniMax H3 or Turbo in the browser. No desktop app, no extension.
A purchase unlocks MiniMax H3 at 768p and Turbo at 720p. We do not sell official 2K, and there is no 2K button that quietly 404s.
Google sign-in and Stripe checkout go through Account. MiniMax H3 credits never expire and cannot move to another website in the matrix.
If Account reports a failed MiniMax H3 generate, credits are not kept for that job. A finished clip can still miss the mark — preview before you post.
“I needed a text to video draft in the browser. One prompt, a 5-second trial, then credits for a longer clip.”
“Image to video is the path I use: drop a product still, describe the move, download the MP4 for the landing hero.”
“When I need 768p or a longer clip I buy credits here and generate in the browser.”
“No app install. This MiniMax H3 AI video generator just runs on my phone, and My works keeps signed-in clips.”
Search MiniMax H3 or Hailuo 3 and you will find MiniMax’s Hailuo app, GitHub weights, and API docs. This site is a separate MiniMax H3 AI video generator with a smaller, honest feature set: 480p until you pay, then 768p in the browser.
MiniMax H3 and Hailuo 3 are one model line. Official apps advertise 2K, omni references, and first/last frames. This MiniMax H3 AI video generator is independent and only runs the text-or-image path we can actually call: 480p until you pay, then 768p.
Paid MiniMax H3 clips here are 768p, 5–15 seconds: text to video, or stills as references. We do not ship official 2K, start/end frames, or lip-sync omni on this page.
Searchers look up GitHub, Hugging Face, ComfyUI, and an official API. This page is a hosted studio, not a weight host. For open H3-Base checkpoints or MiniMax’s own API, use those sites.
How to write a text to video or image to video prompt on this MiniMax H3 AI video generator before you spend credits.
A reliable MiniMax H3 prompt is subject + environment + action + style + camera. One clear action reads cleaner than a packed shot list on a 5–15s clip.
For MiniMax H3 text to video, add a camera verb: slow push in, pan, tracking shot, or locked-off. Example: “red fox in snowy pines, golden hour, cinematic slow motion, gentle tracking shot.”
For MiniMax H3 image to video, do not re-describe the whole still. Name the move: hair in the wind, slow zoom, steam rising. The uploaded frame already holds identity and framing.
MiniMax H3 clips here are 5, 10, or 15 seconds. Buy a pack to generate. A purchase unlocks 768p. We do not run 1080p or official 2K on this page. Failed runs are not charged.
Official MiniMax H3 can take audio references. Here you write the scene (and optionally mention ambience) and download the MP4. There is no lip-sync file picker on this page.
“2K”, “nine reference images”, or “first and last frame” in the prompt will not unlock those official modes. Use the controls that exist: prompt, optional still, duration, aspect.
Buy a pack or subscribe to generate. Go monthly or yearly for the best price per credit — or buy a one-time pack. Credits never expire.
150 credits / month
≈ 30 MiniMax H3 clips / mo at 5s
For the occasional creator.
Cancel anytime
400 credits / month
≈ 80 MiniMax H3 clips / mo at 5s
The plan most creators pick.
Cancel anytime
950 credits / month
≈ 190 MiniMax H3 clips / mo at 5s
For daily, high-volume work.
Cancel anytime
180 credits
$0.083 / credit
A one-time credit top-up.
One-time buy · credits never expire
1,200 credits
$0.050 / credit
Best value one-time buy.
One-time buy · credits never expire
3,000 credits
$0.040 / credit
Bulk credits at the best per-credit price.
One-time buy · credits never expire
Answers
MiniMax H3 is MiniMax’s general-purpose multimodal video model, shipped in 2026 and also written Hailuo 3 or Hailuo 3.0. Official docs say it reads text, images, and optional audio/video context and returns a short clip with picture and 32 kHz stereo in one pass. This site is an independent MiniMax H3 AI video generator for that search: text to video and image to video. A purchase unlocks 768p. We are not MiniMax and we do not run the official 2K / omni Hailuo app here.
Yes. Hailuo is MiniMax’s video product brand, and H3 is the third generation. People search Hailuo 3, Hailuo 3.0, Hailuo H3, MiniMax Hailuo 3, and MiniMax H3 for the same model family. This generator treats Hailuo 3 as the alias.
Hailuo 2.3 was mostly a shorter image-to-video model around 768p or 1080p, often without native stereo. Hailuo 3 / MiniMax H3 is sold as 4–15 seconds at 24 fps, native stereo, a 768p default plus a 2K regenerate, omni references, and first-and-last-frame control. This hosted page runs the text-or-one-still slice: 768p after a purchase.
No. Sign in with Google and buy a credit pack or subscription to generate. There is no guest trial and no signup gift. MiniMax H3 is 768p after a purchase. Failed runs are not charged.
Turbo is our self-hosted video model on this site — not MiniMax H3. After a purchase: Turbo 480p at 4 / 6 / 8 credits for 5s / 10s / 15s, Turbo 720p at 5 / 8 / 10, and MiniMax H3 at 768p for 5 / 8 / 10.
Yes. MiniMax H3 text to video is the default mode on this page: write subject, setting, action, and camera move, then Generate. No footage or editing timeline is required. Pick 16:9, 9:16, or 1:1 before you run.
Yes. MiniMax H3 image to video takes a still (PNG, JPG, or WebP up to 8 MB) as the opening frame. Describe the motion in the prompt. Aspect ratio follows the photo, so the 16:9 / 9:16 picker hides in that mode. Official first-and-last-frame and nine-image omni modes are not on this page.
Not on this site. Official MiniMax H3 / Hailuo 3 can take a first frame, a last frame, or both. Here you may upload one still as the opening frame. If you need a locked landing frame, use MiniMax’s Hailuo app or a host that exposes that mode.
No. Official 2K is a separate in-context regenerate after a 768p take. Omni reference is up to nine images, three videos, and three audio clips. This MiniMax H3 AI video generator sells text to video and image to video at 480p until you pay, then 768p. We do not sell 2K or 1080p.
Use subject + environment + one action + style + camera. Example: “a woman in a yellow raincoat on a rain-slicked street, cinematic lighting, slow push in.” Clips stay short — one beat per generate reads cleaner than a packed shot list.
Upload the still first, then describe only the motion you want: camera move, wind, expression, or lighting change. Do not rewrite the whole scene. If the prompt is empty, this generator uses a simple cinematic-motion default.
On this MiniMax H3 AI video generator you pick 5, 10, or 15 seconds. A purchase unlocks 768p. Official docs also describe a 2K regenerate pass; we do not offer 2K or 1080p here.
MiniMax H3 is built to return native stereo with the picture. This site downloads a plain MP4. We do not take a separate voice file for lip-sync. If you need official omni audio references, use MiniMax’s own app.
No. minimaxh3.lol is independent. We are not affiliated with, endorsed by, or operated by MiniMax, Hailuo AI, or Kuaishou. After a purchase, MiniMax H3 clips on this stack can be 768p. That is still not MiniMax’s own app.
No. Searchers look up an official API, GitHub, Hugging Face, and ComfyUI nodes for open H3-Base weights. This page is a hosted generator, not a weight host and not MiniMax’s API. For weights or the official API, use MiniMax / Hugging Face.
Yes. When a clip finishes it plays in the page and you can download the MP4. Signed-in files also save to My works for about a week.
No. This MiniMax H3 AI video generator runs in the browser on desktop and mobile. There is no app to install and no extension to add.
Each MiniMax H3 generate costs the same credits at either resolution, shown on the button: 5s is 5 credits, 10s is 8, and 15s is 10. 768p unlocks after a purchase. Monthly, yearly, or one-time packs are on Pricing; credits never expire.
Signed-in clips save to My works after you buy and generate. Failed runs are not charged.
Sign in, buy a pack, write a prompt or add a still, and download the MP4.
Generate a video