DeepSeek’s silence has been broken. The company has officially released its new model, DeepSeek 3.1. Built on 685 billion parameters, the model looks set to draw attention with its consistent performance.
The Chinese artificial intelligence startup DeepSeek has officially broken its long-standing silence. The company has officially announced its new model, DeepSeek 3.1. Released as open-source like its predecessors, DeepSeek 3.1 seems poised to make a mark in the artificial intelligence industry with its noteworthy features. Let’s take a closer look at the features of DeepSeek 3.1.
DeepSeek did not provide a detailed statement regarding its new artificial intelligence model. In a post from the company’s official X account, it was stated that the model was trained on 685 billion parameters. The announcement mentioned that the context window has been expanded, but nothing was said about the model’s performance. However…
Independent test results have been revealed
Although official performance tests for DeepSeek 3.1 have not been announced, independent test results have been published. In this context, the model achieved a 71.6 percent success rate in a coding test called Aider, thus becoming one of the best models in the field of coding. The model, which has gained significant user interest, has also become the second most popular model on Hugging Face, according to our checks.
DeepSeek’s new artificial intelligence model aims to provide users with consistent results with a 128k context window. Reviews indicate that the model generally performs well. Of course, we will have to wait for the official announcement to understand these aspects better.
You Might Also Like;
- Mars is Not a Planet. It’s a Graveyard. (The Great Filter Warning)
- Is GTA 6 Proof We’re Living in a Simulation?
- The End of Chatbots: Why 2026 is the Year of AI Agents (And What It Means for Your Job)
Follow us on TWITTER (X) and be instantly informed about the latest developments…
