MiniMax M3

MiniMax M3 is a multimodal foundation model with 512K token context, native multimodal understanding, and MiniMax Sparse Attention (MSA) for efficient long-context inference.

Select Provider