DeepSeek-V3 is the latest model from the DeepSeek team, building upon the instruction following and coding abilities of the previous versions. Pre-trained on nearly 15 trillion tokens, the reported evaluations reveal that the model outperforms other open-source models and rivals leading closed-source models.
For model details, please visit the DeepSeek-V3 repo(opens in new tab) for more information, or see the launch announcement(opens in new tab).
Modalities
In / Out Price
$0.2574 / $1.029per 1M
Context
164K
Released
Dec 26, 2024
Knowledge Cutoff
Jul 2024
DeepSeek-V3 is the latest model from the DeepSeek team, building upon the instruction following and coding abilities of the previous versions. Pre-trained on nearly 15 trillion tokens, the reported evaluations reveal that the model outperforms other open-source models and rivals leading closed-source models.
DeepSeek V3 costs $0.2574/M input tokens and $1.029/M output tokens.
DeepSeek V3 has a 163,840 token context window. It supports up to 16,000 completion tokens.
Yes. DeepSeek V3 accepts tools and tool_choice for function calling. It also supports structured outputs via a JSON schema in response_format.
DeepSeek V3 is served by 3 providers on OpenRouter: StreamLake, DeepInfra and NovitaAI. Requests are routed to the best available provider, with automatic failover to the others, and you can pin or exclude providers with provider routing.
DeepSeek V3 was released on December 26, 2024. Its knowledge cutoff is July 31, 2024.