MiniMax-M1

MiniMax
active

Open-weight hybrid MoE reasoning model with 456B parameters and 1M-token context.

Context window1000K tokens
Input / 1M tokens$0.4
Output / 1M tokens$2.2