Skip to main content
AI Socratic

An open 550B-parameter hybrid Mamba-Transformer MoE (55B active, 1M context). The interesting part is the OpenMDW 1.1 license: full weights plus synthetic data plus training recipes. It's the strongest US open-weights model, and it still trails Kimi K2.6 by ~6 points on the Artificial Analysis Intelligence Index.

Sources: NVIDIA blog, Linux Foundation (OpenMDW)

React:

Comments

Sign in as a member to join the conversation.

Loading comments…

Stay Updated

Get the latest AI insights delivered to your inbox. No spam, unsubscribe anytime.