Chinese AI firm DeepSeek has announced the release of its latest models, V3.2 and V3.2 Speciale, signaling a major push into advanced machine learning. The company claims these systems can compete with offerings from Google and OpenAI, particularly in autonomous reasoning and analytical performance.

V3.2 is designed for general-purpose AI tasks, integrating autonomous tool use for search engines, code execution, and calculations. Benchmarks suggest it performs on par with GPT-5 in logic and reasoning tests, marking a significant step for open-source AI development in China.

Meanwhile, the Speciale version targets complex mathematical and analytical challenges. DeepSeek reports that it performs competitively with Google’s Gemini-3 Pro and demonstrates results comparable to those seen in the International Mathematical Olympiad, highlighting its potential for education and research applications.

The company emphasizes cost efficiency in training the V3 series, claiming it uses fewer resources than comparable Western models. With features optimized for long-document processing and flexible modes for structured reasoning, DeepSeek hopes to make AI more accessible for developers and enterprises.

Read more: https://qazinform.com/news/deepseek-releases-new-ai-systems-challenging-industry-leaders-6fa0fb