StudyArena

MiniMax-M2.1

ReasoningOpen weightsText

MiniMax M-series family model by MiniMax.

About the MiniMax M-series family

MiniMax's open MoE reasoning models (M1 with 40k/80k thinking budgets, M2–M3) using lightning-attention hybrids for long-context efficiency; M2-era models became favorites for agentic coding at low cost.

All MiniMax M-series models

Who makes it

MiniMaxChinaShanghai lab known for efficient long-context M-series models.

More from MiniMax M-series

← Back to all models