In a recent interview, OpenAI CEO Sam Altman praised the new AI model R1 developed by Chinese startup DeepSeek, calling it impressive. However, Altman emphasized that high computing power remains a key success factor for OpenAI.

DeepSeek's Milestones: Reducing AI Training Costs

DeepSeek has garnered public attention with its new model, DeepSeek-V3. A published article revealed that training this model requires less than $6 million in computing power, utilizing less powerful Nvidia H800 chips. These achievements represent a significant step forward in making artificial intelligence more affordable and efficient.

Economic Efficiency of DeepSeek-R1

Last week, DeepSeek unveiled its latest development, the DeepSeek-R1 AI model. According to an official announcement on WeChat, the startup claimed that using the new model is 20 to 50 times cheaper than OpenAI's o1 model, depending on the task. This announcement underscores the company's commitment to making advanced technologies accessible to a broader range of users.

Advantages and Capabilities of DeepSeek-R1

- Cost reduction. Transitioning to the more affordable DeepSeek-R1 can significantly cut down expenses on developing and maintaining AI systems.

- Accessibility of technology. Using less powerful but more accessible Nvidia H800 chips makes AI technology available to small and medium-sized businesses.

- Development potential. Reduced computational costs allow for more funds to be invested in research and development of new technologies.

Computing Power as the Key Success Factor

While DeepSeek shows remarkable achievements in reducing costs, OpenAI remains steadfast in its belief that high computational power is the foundation for achieving superior AI results. According to Sam Altman, more powerful systems not only enhance the accuracy and efficiency of models but also open new horizons for their application.

Key Aspects of High Computational Power

1. Increased model accuracy. Utilizing powerful computational resources enables the creation of more accurate and reliable AI models.

2. Enhanced functionality. High computational power facilitates the development of new features and capabilities for various models.

3. Data processing speed. More powerful systems ensure the swift execution of complex calculations and the handling of large data volumes.

Conclusion

The development and implementation of AI models like DeepSeek-R1 demonstrate significant advancements in reducing costs and increasing the accessibility of AI technologies. Nevertheless, high computational power remains a crucial factor for achieving maximum efficiency and precision of models. The contrasting approaches and successes of both DeepSeek and OpenAI highlight the diversity of strategies and perspectives in the rapidly evolving field of artificial intelligence.