AI Release Timeline

DeepSeek V4 launched by DeepSeek. Next-gen DeepSeek general model emphasizing coding and cost efficiency.

API endpoint `deepseek-v4` available for DeepSeek V4.

Initial pricing set at $0.30/M input and $1.20/M output tokens.

DeepSeek V4 supports function calling and structured tool use.

DeepSeek V4 introduces extended reasoning and chain-of-thought capabilities.

DeepSeek V4 achieves 92/100 coding benchmark score — top-tier performance.

DeepSeek V4 available for fine-tuning and custom deployment via DeepSeek.

DeepSeek V3 0324 launched by DeepSeek. Updated V3 with improved instruction following and 128K context.

API endpoint `deepseek-chat` available for DeepSeek V3 0324.

Initial pricing set at $0.27/M input and $1.10/M output tokens.

DeepSeek V3 0324 supports function calling and structured tool use.

DeepSeek V3 0324 achieves 92/100 coding benchmark score — top-tier performance.

DeepSeek V3 0324 available for fine-tuning and custom deployment via DeepSeek.

DeepSeek R1 launched by DeepSeek. Open reasoning model with o1-class performance at fraction of cost.

API endpoint `deepseek-reasoner` available for DeepSeek R1.

Initial pricing set at $0.55/M input and $2.19/M output tokens.

DeepSeek R1 supports function calling and structured tool use.

DeepSeek R1 introduces extended reasoning and chain-of-thought capabilities.

DeepSeek R1 achieves 95/100 reasoning benchmark score — top-tier performance.

DeepSeek R1 available for fine-tuning and custom deployment via DeepSeek.

DeepSeek V3 launched by DeepSeek. Highly efficient MoE model rivaling frontier closed models.

API endpoint `deepseek-chat` available for DeepSeek V3.

Initial pricing set at $0.27/M input and $1.10/M output tokens.

DeepSeek V3 supports function calling and structured tool use.

DeepSeek V3 available for fine-tuning and custom deployment via DeepSeek.

DeepSeek Coder V2 launched by DeepSeek. Code-specialized DeepSeek for programming tasks.

API endpoint `deepseek-coder` available for DeepSeek Coder V2.

Initial pricing set at $0.14/M input and $0.28/M output tokens.

DeepSeek Coder V2 supports function calling and structured tool use.

DeepSeek Coder V2 achieves 93/100 coding benchmark score — top-tier performance.

DeepSeek Coder V2 available for fine-tuning and custom deployment via DeepSeek.