MiniMax H3 Max — Fast Text and Image to Video

MiniMax H3 Max is the speed-tuned version of MiniMax H3. Type a prompt or drop in a still, and a 5-15 second clip comes back in seconds rather than minutes. Pay per second of finished video, with no plan to sign up for.

• Fast generation: Tuned for turnaround. Short clips come back in seconds, changing prompting from a batch job into a conversation.
• 480P & 768P resolutions: Choose between a lower resolution for drafts or a higher resolution for publishing.
• 5-15 second clips: Generate clips of any whole second duration within this range, ideal for social media and short-form content.
• Text and image input: Create videos from text prompts or animate still images by providing a first frame, a last frame, or both.
• Pay per second: Billing is based on the output duration. Input images are free, and failed generations are refunded.
• No subscription required: Pay-as-you-go model means you only pay for the video seconds you generate.

What you can build with this model:

• Short-form character video, at volume: Generate consistent character videos for social media content efficiently.
• Title cards and motion design: Create dynamic title sequences and logo reveals by providing a starting frame.
• Narrative scenes you can iterate on: Quickly generate and refine short narrative scenes due to the fast turnaround time.

Comparison with MiniMax H3:

• MiniMax H3 Max offers faster generation speeds and is optimized for high-volume short video production.
• It supports 480P and 768P resolutions, while MiniMax H3 supports up to 2K.
• MiniMax H3 Max has a clip length of 5 to 15 seconds, whereas MiniMax H3 supports 4 to 15 seconds.
• Crucially, MiniMax H3 Max does not support reference images, video, or audio inputs, a feature retained by MiniMax H3 for more complex conditioning.

API Details:

The MiniMax H3 Max model is accessible via the reAPI video endpoint. Integrations can switch between MiniMax H3 Max and MiniMax H3 by changing a single parameter in their API requests.