DeepSeekV3.2 leads the open source agent evaluation

DeepSeek releases V3.2 and V3.2-Speciale: Inference-first models for agent scenarios

Two months ago, DeepSeek released its experimental model V3.2-Exp and confirmed that it was not inferior to V3.1-Terminus in various scenarios through...