Minimax-H3

Model IntroductionText

MiniMax H3 is a general-purpose, omni-modal generative system.

Introduction

MiniMax H3 is a general-purpose, omni-modal generative system. It supports unified understanding of multimodal contexts composed of text, images, video, and audio, and can generate video with native stereo audio at resolutions up to 2K and durations of up to 15 seconds. Thanks to its task-generalization-oriented system design, H3 already possesses broad multimodal context understanding and generation capabilities at the pre-training stage, enabling outstanding performance in following complex multimodal instructions.

Key Features

  • H3-Base encodes different modalities using their corresponding encoders or VAEs and organizes the encoded representations into a unified packed multimodal sequence. RoPE is used to capture the necessary spatial and temporal relationships among tokens before the entire sequence is passed to the H3-Omni-Transformer.
  • Specifically, text is encoded by the H3-Encoder; visual inputs are encoded by both the H3-Encoder and the H3-VisualVAE; and audio is encoded solely by the H3-AudioVAE.
  • The H3-Omni-Transformer jointly predicts video and audio latents, which are then decoded into video and stereo audio, respectively.
  • To reduce the computational cost of long multimodal sequences, H3 natively supports sparse-attention training and inference. The initial open-source release provides inference with full attention only. Our sparse-attention implementation will be released in a future update.

Benchmarks

Benchmark (Area)Score*
HappyWorld-Bench (W1: Appearance Quality & Consistency) (Overall)81.62 (Eval)
Physical World Reasoning (Omni-Modal) (Overall)41.79% (Eval)
Professional Editing-Technique Execution (Overall)0.617 (Eval)
Artificial Analysis Video Arena (Overall)1247.61 (Arena)
Arena (Public Leaderboard) (Image-to-Video)1476 ±19 (Arena)

API request example

Endpoint /v1/chat/completions. Replace <API_KEY> with a key from the console.

Standard OpenAI Chat Completions. Swap `model` for the target slug.

curl https://api.ravenrouter.ai/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <API_KEY>" \
  -d '{
  "model": "Minimax-H3",
  "messages": [
    {
      "role": "user",
      "content": "Hello! Introduce yourself in one sentence."
    }
  ],
  "max_tokens": 64
}'

More models: API Reference → Examples by model

Minimax-H3 API — $0.13/1M input, $0.32/1M output | RavenRouter AI