MiniMax M3 is a multimodal foundation model with 1M token context, native multimodal understanding, and MiniMax Sparse Attention (MSA) for efficient long-context inference.
LLM Gateway routes requests to the best providers that are able to handle your prompt size and parameters.