MiniMax has launched MiniMax M3, an open-weight model featuring a one million token context window and native multimodal training from scratch. It uses a custom sparse attention system to achieve up to fifteen times faster processing.
FireBrief
FireBrief