On August 21, DeepSeek Company announced on its WeChat official account that it has officially released DeepSeek-V3.1.
The models used by the official web page, App, mini program, and API platform have been updated synchronously. The new model identifies itself as DeepSeek-V3.
DeepSeek introduced that DeepSeek-V3.1 uses the parameter precision of UE8M0 FP8 Scale. In the comment section, this account stated in a top comment: "UE8M0 FP8 is designed for the next-generation domestic chip to be released."
This upgrade includes the following main changes: Hybrid inference architecture: one model supports both thinking mode and non-thinking mode; higher thinking efficiency: compared to DeepSeek-R1-0528, DeepSeek-V3.1-Think can provide answers in a shorter time; stronger Agent capabilities: through Post-Training optimization, the new model shows significant improvements in tool usage and agent tasks.
The official App and web version models have been updated to DeepSeek-V3.1. Users can switch freely between thinking mode and non-thinking mode by clicking the "Deep Thinking" button.
Original article: https://www.toutiao.com/article/7541004477101294119/
Statement: This article represents the views of the author. Please express your opinion by clicking the [Like/Dislike] buttons below.