MiniMax-M3开源多模态模型登陆Together AI，支持100万token上下文
英文摘要
MiniMax-M3, an open-weight native multimodal model from MiniMax, is now available on Together AI, the company’s preferred cloud partner. The model features a 1 million token context window, MiniMax Sparse Attention for efficiency, and supports both thinking and non-thinking inference modes. Together AI has optimized inference for MiniMax-M3, achieving up to 125% higher throughput across various concurrency levels, making the model accessible with enhanced performance.
中文摘要
MiniMax公司发布的原生多模态模型MiniMax-M3现已在Together AI平台上线，后者是其首选云合作伙伴。该模型为开放权重，拥有100万token上下文窗口，采用MiniMax稀疏注意力机制，并支持思考与非思考两种推理模式。Together AI针对该模型进行了推理优化，在不同并发水平下吞吐量最高提升125%，使开发者能够以更高性能访问该模型。
关键要点
MiniMax-M3 is an open-weight native multimodal model now available on Together AI.
MiniMax-M3是一款开放权重的原生多模态模型，现已在Together AI上线。
It features a 1M token context window, MiniMax Sparse Attention, and thinking/non-thinking modes.
该模型拥有100万token上下文窗口、MiniMax稀疏注意力机制，以及思考与非思考模式。
Together AI is MiniMax’s preferred cloud partner and provides inference optimizations yielding up to 125% higher throughput.
Together AI是MiniMax的首选云合作伙伴，其推理优化可将吞吐量提升高达125%。