AI Architecture & Systems

Model Family: DeepSeek

6.30.1DeepSeek LLM and DeepSeekMoE#

6.30.2DeepSeek-V2: MLA and Fine-Grained MoE#

6.30.3DeepSeek-V3: Low-Precision Training, Load Balancing, and MTP#

6.30.4DeepSeek-R1: Reasoning RL and Distillation#

6.30.5DeepSeek-V3.1/V3.2: Reasoning Modes and Sparse Attention#

6.30.6DeepSeek-V4: Public Model Report and Deployment Requirements#

6.30.7Model–Kernel–System Co-Design Across the DeepSeek Family#