New Achievements in Artificial Intelligence in May 2025

Like everyone else, we literally could barely keep up with the AI news feed. May was overflowing with news, we're keeping the main ones in our blog for memory.
Summary:
Breakthroughs in Generative AI and Hardware Solutions
- OpenAI, together with Jony Ive, announced the development of a new AI device of the future, which could become a revolution in consumer electronics and AI integration into everyday life.
- China launched a supercomputer for AI computing into orbit, opening new horizons for big data processing and autonomous AI systems in space.
DeepSeek Released Updated R1 Model with 685 Billion Parameters
- Chinese company DeepSeek presented an improved version of its R1 model, increasing the number of parameters to 685 billion, which enhances its performance and accuracy.
AI in Medicine and Healthcare
- China opened the first "AI hospital" where artificial intelligence serves up to 400 thousand patients, optimizing diagnosis and treatment while reducing the burden on medical staff.
- In healthcare, AI models are being used for personalized medicine, early diagnosis, and creating individual treatment plans, significantly improving the quality of medical care.
Innovations in Education
- The Higher School of Economics (HSE) received the AI Awards 2025 for implementing AI in educational processes — an automated assignment checking system reduced evaluation time by 40% and increased objectivity.
Development of Generative and Business-Oriented AI Tools
- May 2025 marked the growth in number and quality of AI-based business tools: automated video creation, text generation, improved voice-overs, smart chatbots, and analytics. These solutions significantly enhance productivity and personalization of business processes.
- Generative AI is becoming increasingly accessible, allowing companies and individuals to create complex content and visualizations without specialized knowledge.
Quantum AI and Computing Power
- The field of quantum artificial intelligence is actively developing, enabling solutions to problems inaccessible to classical computing systems. The integration of quantum computing into AI promises a revolution in big data processing and optimization of logistics, financial, and scientific tasks.
Google Introduced Veo 3 — Video Generation Model with Audio Support and 4K Resolution
- In May 2025, Google announced the Veo 3 model, capable of generating videos with up to 4K resolution and supporting audio synthesis. This is a significant step in the development of multimedia content generation technologies.
Narrowing the Gap Between Open and Closed AI Models
- The performance gap between advanced closed AI models (such as GPT-4) and the best open solutions has narrowed to 1.7%, making cutting-edge technologies more accessible to a wider audience.
Growth in Computing Power and Reduction in AI Costs
- AI chip speed grows annually by 43%, and energy efficiency by 40%, while computing costs fall by 30% per year. This makes AI technologies more accessible for mass adoption.
Overall, May 2025 was marked by large-scale implementations of AI in medicine, education, business, and everyday life, as well as technological leaps in generative, quantum, and responsible AI systems.