Minimax-01
Introduction
Advanced AI models developed by MiniMax with large context windows and efficient architecture
Added on
Apr 07 2025
Features
Minimax-01 AI Information
What is Minimax-01?
MiniMax-01 refers to a series of AI models developed by MiniMax, a Chinese AI lab. These models are notable for their large context windows (up to 4 million tokens) and efficient hybrid architecture combining Lightning Attention, Softmax Attention, and Mixture-of-Experts (MoE). The series includes text, visual, video and speech models with commercial-grade quality.
How to use Minimax-01?
You can access MiniMax-01 through API (Hailuo AI platform or OpenRouter), open source models on GitHub/Hugging Face, or cloud deployment options. For video generation, use text descriptions or images as prompts. The Director model allows natural language commands to control camera movements. Pricing is $0.2 per million input tokens and $1.1 per million output tokens.
Minimax-01's Core Features
- 4 million token context window for processing entire documents
- Hybrid architecture with Lightning Attention for efficiency
- Mixture of Experts (MoE) with 456B total parameters
- Multimodal capabilities (text, image, video, speech)
- Cost-effective compared to GPT-4o and Claude 3.5