MiniMax/MiniMax-H3

MiniMax H3 是一个通用型全模态生成系统。它能够统一理解由文本、图像、视频和音频构成的多模态上下文,并可生成最高 2K 分辨率、最长 15 秒、带有原生立体声音频的视频。得益于以任务泛化为导向的系统设计,H3 在预训练阶段便已具备广泛的多模态上下文理解与生成能力,因此能够出色地遵循复杂的多模态指令。

API model names

Use any of these names as the model parameter in API requests.

Price
¥0.5 • ¥0.5
Cache Price
¥0.5 • ¥0.5
Input
Output

Start using MiniMax/MiniMax-H3 now

Support Subscription & Pay-as-you-go matching your needs in Studio & API

View Pricing

Specifications

Context Window
4K
Max Output Tokens
4K

Modalities

Text

Input and output

Image

Input only

Audio

Input only

Video

Input and output

Endpoints

Core Generation
Chat Completions
/v1/chat/completions
Responses
/v1/responses
Media Generation
Embeddings
/v1/embeddings
Image Generation
/v1/images/generations
Image Edit
/v1/images/edits
Videos
/v1/videos
Extended Protocols
Messages
/v1/messages
Generate Content
/v1beta/models/{model}:generateContent

Ready to use MiniMax/MiniMax-H3?

Sign up for WisGate to access MiniMax/MiniMax-H3 in our visual Studio or seamlessly integrate via API

Subscription & Pay-as-you-go
Powerful API Access
High Cost-Effectiveness
Interactive Studio UI
View Pricing

Latest Models

Explore the newest AI models just arrived

DeepSeek V4 Pro

DeepSeek•10/26/2026

MiniMax/MiniMax-H3

Minimax•10/23/2026

GLM-5.3

Z-AI•10/23/2026

DeepSeek-V4.1-Flash-1017

DeepSeek•10/23/2026

Qwen-Image-2.1-Pro

Qwen•10/23/2026

Qwen3.8-Omni-Flash-Realtime

Qwen•10/23/2026