All tracked DeepSeek models — pricing, context windows, lifecycle state, and release history in one place.
DeepSeek V3.2 thinking mode via deepseek-reasoner endpoint. Up to 64K output with reasoning chains.
DeepSeek's latest model via deepseek-chat endpoint. V3.2 non-thinking mode. 128K context.
Open-weight reasoning model. Chain-of-thought with distilled variants. Superseded by V3.2 thinking.
Open-weight MoE model. 671B total, 37B active params. Superseded by V3.2 on API.
Previous-generation MoE model. 236B total params. Superseded by V3.
Predecessor → successor chains tracked for DeepSeek models.
DeepSeek's latest model via deepseek-chat endpoint. V3.2 non-thinking mode. 128K context.
DeepSeek V3.2 thinking mode via deepseek-reasoner endpoint. Up to 64K output with reasoning chains.
Open-weight reasoning model. Chain-of-thought with distilled variants. Superseded by V3.2 thinking.
Open-weight MoE model. 671B total, 37B active params. Superseded by V3.2 on API.
Previous-generation MoE model. 236B total params. Superseded by V3.