
DeepSeek introduced V4.1-Flash for long-context processing
The multimodal model DeepSeek-V4.1-Flash activates 8 billion parameters when processing input and 16 billion during generation. Weights are published under the MIT license.
Read the newsWhat's new in IT, AI, and science.
Briefly about what's interesting.

The multimodal model DeepSeek-V4.1-Flash activates 8 billion parameters when processing input and 16 billion during generation. Weights are published under the MIT license.
Read the news
The company claims that its AI system has proven the possibility of breaking smooth flow in a three-dimensional incompressible fluid. The proof is posted along with a formalization in Lean,…
Read the newsDeepMind's new model issues forecasts every hour with resolution up to 5 km and predicts precipitation more accurately. It has already been integrated into Search, Maps, and Gemini.
Read the news
The history of the rivalry between Marvin Minsky and Frank Rosenblatt, who knew each other since school: in 1969, Minsky and Papert published a book showing the limitations of the…
Read the news
OpenAI has announced GPT-6 Astra. The new model will first become available to a limited set of organizations, and then to ChatGPT Plus, Pro, Business, and Enterprise subscribers, as well…
Read the news