dots3-note-preview

280B Model, 16B Active: Open-Source AI Shifts From Total to Active Parameters

dots3-note preview: 280B MoE, 16B active, 512K context, multimodal. Open-source AI race now measures compute cost, not parameter count.