Back to section
Event · 2025

Open models

DeepSeek-V3 and R1

Strong reasoning at a radically lower training cost.

Sources
DeepSeek-V3 · technical reportOpen primary source