MiniMax H3 Max Text to Video API

MiniMax H3 Max Text to Video

MiniMax H3 Max Text to Video is built for high-throughput video production, combining generation faster than real time with precise prompt adherence. It natively synthesizes synchronized stereo audio and video with physical action across multiple shots. With 5–15 second durations, resolutions from 480p to 1080p, and multiple aspect ratios, it offers cost-effective production for film previs, animated concept shorts, and creative advertising.

Why choose the MiniMax H3 Max Text to Video API?

• High-throughput generation faster than real timeAdvanced post-training optimization and accelerated inference can render a 5-second video in about 3 seconds under ideal conditions, shortening creative validation and batch-production cycles.

• Precise adherence to complex promptsUnderstands subject features, action choreography, shot-size changes, and lighting instructions, with substantially improved fidelity to long, multi-stage descriptions compared with the base model.

• Native synchronized stereo audio and videoGenerate ambient effects, action sounds, and atmospheric soundscapes aligned with the visual rhythm in a single pass, without adding separate audio in post-production.

• Six standard aspect ratios for commercial useNatively supports six aspect ratios, from 21:9 ultrawide and 16:9 landscape to 9:16 vertical shorts, covering film previs and major social platforms.

• Flexible durations with transparent per-second billingChoose any whole-second duration from 5–15 seconds and 480p, 768p, or 1080p resolution. Billing follows the requested duration parameter for predictable budgeting.

Parameters

How to Use

1. 01

   Describe the subject and settingDefine the main subject, appearance, clothing textures, time and place, weather, and lighting.

2. 02

   Plan the main action and camera workUse clear verbs to describe character or object movement, and specify camera pushes, pulls, pans, tracking paths, and shot-size changes.

3. 03

   Add synchronized sound cuesInclude descriptions of ambient sound, impact effects, or background audio to guide matching native stereo sound.

4. 04

   Set output specifications and framingChoose a whole-second duration from 5–15 seconds, 480p/768p/1080p resolution, and an aspect ratio suited to your delivery channel.

5. 05

   Submit the task and retrieve the videoSubmit a generation request through the API to receive a task ID, then poll the status endpoint or configure a callback URL to receive the completed MP4 file.

Usage and billing

The MiniMax H3 Max Text to Video API uses transparent per-second pricing by resolution tier. Cost depends only on the requested duration parameter and selected resolution, with no hidden conditions.

Best Use Cases

• Film and animation previsQuickly turn written scripts into animated storyboards with sound to assess camera rhythm, blocking, and dramatic impact.

• Frequent social media shortsBatch-generate topical shorts and ad concepts in 9:16 or 16:9 to reduce production costs.

• Ad concepts and visual explorationTry different art styles, lighting, and narrative ideas early in a project, then quickly share concept prototypes with clients and teams.

• Game cinematics and environment previewsQuickly create moving landscapes, sci-fi rainy nights, or complex city scenes for game backgrounds, atmosphere previews, and promotional assets.

Pro Tips

• Structure your prompt in sections: start with the main subject and appearance, then the action sequence, camera placement and movement, and finally lighting and color style.
• Describe sound effects clearly: add specific sounds at the end of the prompt, such as footsteps, rain hitting a car window, or train wheels striking the rails, to improve audiovisual alignment.
• Keep each shot focused on one main event: for short 5–8 second shots, build around a coherent action rather than unrelated characters or abrupt changes of scene.
• Match composition to aspect ratio: for 21:9 widescreen or 9:16 vertical output, emphasize scene depth or vertical composition in the prompt accordingly.
• Validate quickly at 768p: use 768p to generate and compare iterations at scale, then render the selected shot at 1080p for final delivery.

Notes

• The MiniMax H3 Max Text to Video API accepts text prompts only. Source images, reference video, and audio inputs are not supported.
• Successful submission immediately returns a task_id. Tasks run asynchronously, and the video can be downloaded only after generation finishes.
• Poll the status endpoint every 2 seconds while the task is not_started or running. Stop once it reaches finished or failed.
• Pass callback_url at the request's top level to automatically receive an asynchronous Webhook notification when generation finishes or ends in an error.