DeepSeek releases upgraded AI models that blend reasoning with autonomous tool use

DeepSeek has introduced two updated versions of its experimental AI model, marking a significant step in its effort to blend advanced reasoning with autonomous action. The company announced on X that its new DeepSeek-V3.2 model integrates “thinking” directly into tool use, enabling it to operate in both reasoning and non-reasoning modes. Bloomberg reported that these versions expand on models released only weeks earlier, adding capabilities meant to combine logical processing with the ability to carry out tasks independently.

According to DeepSeek, the latest models match the performance of OpenAI’s GPT-5 on several reasoning benchmarks, a sign that China’s open-source AI platforms continue to compete closely with major U.S. offerings. The V3.2 model is designed to mimic forms of human-like reasoning while using resources such as search engines, calculators, and code executors to complete tasks.

DeepSeek drew global attention early this year when it released an AI system that it said could rival leading U.S. models at a fraction of their development cost. Industry observers noted that this approach challenged assumptions about the high financial and environmental burdens of cutting-edge AI. SAP consultant Gokul Naidu said the launch highlighted an important shift, showing that strong AI performance and affordability can coexist. He added that this could make advanced AI more accessible to small and midsize companies instead of remaining limited to large tech giants.

PYMNTS also recently reported on how enterprise attitudes toward AI have changed three years after the launch of ChatGPT. Research from PYMNTS Intelligence shows that many companies now treat AI as a core part of their operations. Sixty major U.S. firms have installed chief AI officers as they reorganize responsibilities around the technology. Increased output was cited by 34% of CFOs as the leading reason for adopting AI, followed by competitiveness and improved decision-making. The study noted clear industry differences, with goods-sector companies focused on boosting efficiency, while service sectors aim to enhance customer experience. Technology firms, meanwhile, are prioritizing AI to maintain their competitive edge.