MiniMax H3 Max Turbo Text to Video API

MiniMax H3 Max Turbo Text to Video

MiniMax H3 Max Turbo Text to Video uses a distilled, post-trained flagship architecture to increase throughput and shorten generation time. It accepts prompts of 1–50,000 characters and supports 480p, 768p, and 1080p output in six aspect ratios, with native synchronized audio and video. Per-second pricing starts at 5 credits, supporting frequent iteration for ad previews and social shorts.

Why choose the MiniMax H3 Max Turbo Text to Video API?

• Fast generation at high throughputPost-training distillation and optimized inference significantly improve speed over the base model, reducing wait times for fast creative exploration and batch content production.

• Native synchronized stereo audio and videoVisuals and ambient effects render together, producing a complete audiovisual short without a separate audio prompt or an added soundtrack in post-production.

• Flexible framing for every formatSix standard aspect ratios, 21:9, 16:9, 4:3, 1:1, 3:4, and 9:16, cover widescreen storytelling, traditional formats, and vertical social content for mobile.

• Flexible duration and resolution tiersChoose a whole-second duration from 5–15 seconds and 480p, 768p, or 1080p resolution to balance rendering speed and visual detail.

• Per-second billing at half the costRates are just 50% of the standard flagship version, starting at 5 credits per second ($0.025/second) to make commercial video production more affordable.

Parameters

How to Use

1. 01

   Define the subject and settingStart with a clear natural-language description of the main subject, surroundings, and background details to establish the visual tone.

2. 02

   Plan the main actionUse precise verbs to describe the scene's central action, keeping the short video focused on one coherent, expressive moment.

3. 03

   Specify camera work and lighting rhythmAdd shot size and angle, camera paths such as pans, push-ins, pullbacks, or tracking, plus lighting atmosphere and color style.

4. 04

   Set aspect ratio and output durationChoose an aspect ratio for your publishing platform, then select a whole-second duration from 5–15 seconds and a resolution tier suited to the action's complexity.

5. 05

   Submit the task and retrieve the resultSubmit a generation request to receive a task_id, then poll the status endpoint or configure a callback URL to retrieve the rendered video.

Usage and billing

Cost depends on the requested duration parameter and selected resolution tier, starting at 5 credits per second. Rates are fixed for each tier, with transparent per-second billing and a live cost estimate in the dashboard before submission.

Best Use Cases

• Commercial marketing and concept previewsQuickly turn creative scripts into expressive audiovisual video drafts for team pitches and client reviews.

• Social media and short-video campaignsUse 9:16 and 1:1 framing to quickly produce dynamic social content and promotional assets designed for conversion at scale.

• Film and game shot previewsExplore blocking, composition, perspective, motion rhythm, and lighting style before filming or 3D production.

• Visual effects and atmosphere testsQuickly test complex physical motion, particle effects, fireworks, or animated lighting at low cost to refine an art direction.

Pro Tips

• Structure prompts as "subject and setting -> subject action -> camera path -> lighting and style" to help the model understand the intended scene.
• Focus each short video on the progression of one main action. Avoid stacking too many separate, complex instructions into a single prompt.
• Specify subject and camera directions separately, such as "the subject runs forward while the camera tracks backward from a low angle."
• Choose the target aspect ratio before planning the framing so subject placement, negative space, and composition match the final output.
• To emphasize a sound effect, describe the environment or action that produces it, such as rain tapping on glass or a vehicle speeding past.

Notes

• MiniMax H3 Max Turbo Text to Video accepts text only. Requests containing source-media fields such as images or videos will be rejected.
• Successful submission immediately returns an asynchronous task_id rather than the finished video file.
• Poll every 2 seconds while status is not_started or running. Stop when the task reaches finished or failed.
• 1080p output uses latent refinement upsampling. For maximum inference efficiency and value, the native 768p tier is recommended.
• Pass callback_url at the request's top level to automatically receive a Webhook notification containing the video download URL when the task finishes.