A new-generation open general-purpose multimodal video model. It understands creative intent across multimodal context — text, image, video, and audio — and delivers more natural, coherent generation and expression.
Released MiniMax-M2.5 series models (MiniMax-M2.5 / M2.5-highspeed). Achieving or setting new SOTA benchmarks in programming, tool calling and search, office productivity and other scenarios.
Released the Speech-02 model series, delivering hyper-realistic vocal performance. The models are distinguished by exceptional prosody and stability, along with outstanding voice similarity and audio quality.
Released a video generation model with enhanced camera control, offering superior adherence to camera movement directives and enabling cinematic storytelling through its shot composition.