The recent unveiling of the Gemini2.5Pro model by Google DeepMind has captured the attention of the tech world. As one of the leading AI large language models, Gemini2.5Pro demonstrates unprecedented application potential thanks to its ability to handle million-scale long contexts. However, despite its technological edge, the model's operation remains costly, and there is still room for quality improvement.

The Gemini series stands out for its ability to process ultra-long contexts, a feature that makes it particularly effective in AI programming and information retrieval. Compared to other models, Gemini2.5Pro can read the entire content of a project at once, providing a smoother and more efficient user experience. This technology marks a new phase for large models, potentially transforming traditional information interaction methods.

In a conversation with podcast host Logan Kilpatrick, Google DeepMind research scientist Nikolay Savinov emphasized the importance of context. He noted that the context information provided by users can greatly enhance the model's personalization and accuracy. The model relies not only on pre-trained knowledge but also on real-time user inputs to update and adjust its responses, ensuring the timeliness and relevance of the information.

Savinov also mentioned that RAG (Retrieval-Augmented Generation) technology will not be phased out but will work in tandem with long contexts. This technology helps the model quickly retrieve relevant information from vast knowledge bases through a pre-processing step, further enhancing the recall rate of information based on million-scale contexts. The combination of the two can significantly improve the model's performance in practical applications.

The future outlook for long context technology is also optimistic. As costs gradually decrease, it is expected that the ability to handle contexts of tens of millions will become the industry standard in the near future. This will undoubtedly bring revolutionary breakthroughs in AI coding and other application scenarios.

Gemini2.5Pro not only drives the development of AI technology but also opens up new possibilities for enhancing user experience. The application of long contexts and their combination with RAG technology herald a future where AI will become more intelligent and personalized.