The MiniMax M2 is here. It is an exciting model that has just achieved an all-time-high index score for open weight models. It has 200B total parameters but is very sparse with only 10B active parameters per forward pass. This model can fit on 4xH100s at FP8 precision. This model can follow instructions better and use tools.
MiniMax’s M2 achieves a new all-time-high Intelligence Index score for an open weights model and offers impressive efficiency with only 10B active parameters (200B total)
Key takeaways:
➤ Efficiency to serve at scale: MiniMax-M2 has 200B total parameters and is very sparse with… pic.twitter.com/Z2ktn6s83z— Artificial Analysis (@ArtificialAnlys) October 27, 2025
As Artificial Analysis on X explains, MiniMax’s API costs 0.3/$1.2 per 1M input/output tokens. More info is available here.
