Skip to main content
Our models seamlessly combine text, speech, and vision, each with support for hundreds of billions of parameters. We optimize the entire product lifecycle, from research to deployment. Based on our expansive experimental and inference platforms, we’re able to accelerate technological and product iterations, continuously enhancing both model performance and user experience. Our inference platform is powered by an ultra-efficient multi-cloud computing system, leveraging tens of thousands of GPUs to support over a billion API calls per day. Our general-purpose models are regularly updated and offer powerful capabilities in text, image, and audio processing. Meanwhile, our AI engine’s training platform harnesses highly efficient parallel computing with thousands of GPUs, ensuring exceptional stability and performance. We provide both enterprise and independent developers with feature-rich and user-friendly APIs that reduce development complexity. This allows users to quickly validate their solutions in target scenarios and proceed to production deployment. Contact us: api@minimax.io

​
Models and Capabilities

  • Text model ( MiniMax-M2 ) : New Reasoning Model, Global Top-Tier Performance, Ultra-Long Context, Ultimate Value.
  • Speech model ( Speech-2.5-preview ) : A high-intelligent model that converts text to speech based on natural language interaction.
  • Video model ( MiniMax Hailuo 02) : Supports the generation of high-definition videos at 1080p resolution, and max 10s duration.

​
Application and Use-cases

  • Voice model key features include:
Text-to-speech, Rapid and high-quality voice cloning, Multi-language, multi-character voice generation.Applications include social media, podcasts, audiobooks, news, education, VR personas, and more.

​
Why Choose MiniMax Models

  • High Performance：Our trillion-parameter language model supports over a billion daily requests with a robust compute pool of 10,000 GPUs. Regular large-scale training sessions and weekly updates on the self-developed training framework ensure cutting-edge performance.
  • Versatile & User-Friendly: With a wide array of customizable parameters and flexible usage methods, MiniMax offers countless application scenarios.
  • Concurrency: Extensive inference clusters effortlessly handle high-volume user applications.
  • Robust Security: Safeguard your data’s security with dual-layer protections, combining advanced language model security features with independent third-party audits.
  • Expertise & Innovation: Leading R&D and business teams deliver cutting-edge AGI technology services and solutions.
  • Enter a name for the key in the provided field, then click Create.