MiniMax H3
MiniMax H3 is a video generation model from MiniMax that supports text-to-video, image-to-video, and reference-based video creation at up to 2K resolution.
MiniMax
WaveSpeedText and image to 2K video generation
MiniMax H3 is a video generation model developed by MiniMax and made available through the WaveSpeed provider on MindStudio. It supports multiple generation modes including text-to-video, image-to-video, and reference-to-video, allowing users to drive video output from written prompts, a first or last frame image, or a set of reference images, videos, and audio clips. The model was released in July 2026 and is priced between $0.09 and $0.13 per second of generated video.
H3 is designed for workflows that require flexible input combinations and high-resolution output, with support for up to 2K resolution and configurable aspect ratios and durations. The reference-to-video capability allows users to supply reference images, videos, and audio as conditioning inputs, giving creators more control over the visual style and content of the output. It is well suited for content creators, product teams, and developers building video generation pipelines who need a model that accepts diverse multimodal inputs without requiring separate preprocessing steps.
What MiniMax H3 supports
Text to Video
Generates video clips directly from a text prompt, using the written description to determine scene content, motion, and composition.
Image to Video
Animates a provided image into a video clip, accepting both a first frame and a last frame image to guide the motion between two visual endpoints.
Reference-Based Generation
Accepts reference images, videos, and audio as conditioning inputs to influence the style, subject, or audio characteristics of the generated video.
2K Resolution Output
Supports video generation at up to 2K resolution, with a selectable resolution toggle and configurable aspect ratio settings.
Configurable Duration
Allows users to select the output video duration via a dedicated input, giving control over clip length within the model's supported range.
Aspect Ratio Control
Provides a selectable aspect ratio input so generated videos can be formatted for different platforms or display contexts.
Ready to build with MiniMax H3?
Get Started FreeCommon questions about MiniMax H3
What generation modes does MiniMax H3 support?
MiniMax H3 supports text-to-video, image-to-video, and reference-to-video modes. Users can provide a text prompt, a first or last frame image, or a combination of reference images, videos, and audio to condition the output.
What is the pricing for MiniMax H3?
MiniMax H3 is priced between $0.09 and $0.13 per second of generated video, depending on the selected settings such as resolution and duration.
What is the context window for MiniMax H3?
MiniMax H3 has a context window of 2,000 tokens, which applies to the text prompt input used to describe the desired video content.
What resolution does MiniMax H3 support?
MiniMax H3 supports video generation at up to 2K resolution. Users can select the resolution and aspect ratio through dedicated input controls.
When was MiniMax H3 released?
MiniMax H3 was released in July 2026 and became available on MindStudio on August 3, 2026.
Parameters & options
First-frame image URL. Supported image dimensions are 256-5760 pixels per side, with an aspect ratio between 0.4 and 2.5.
Optional last-frame image URL.
Reference image URLs. At least one reference image or video is required.
Reference video URLs. At least one reference image or video is required.
Optional reference audio URLs. Audio cannot be provided alone.
Up to 3 LoRAs. 2K is not available when LoRAs are applied.
Output video duration in seconds.
Explore similar models
MiniMaxHailuo 2.3 Pro
Hailuo 2.3 Pro is a video generation model from MiniMax that supports both text-to-video and image-to-video creation.
MiniMaxMiniMax Music 3.0
MiniMax Music 3.0 is a music generation model from MiniMax that creates songs from lyrics or as instrumentals at $0.15 per song.
MiniMaxMiniMax H3 Image
ByteDanceSeedance 2.5 Turbo
Seedance 2.5 Turbo is a video generation model from ByteDance that supports text-to-video, image-to-video, and video editing modes.
LightricksLTX-2.5
WanWan 3.0
Wan 3.0 is a video generation model that converts text prompts and images into video clips with optional audio output.
Start building with MiniMax H3
No API keys required. Create AI-powered workflows with MiniMax H3 in minutes — free.