Translated data: Microsoft has open-sourced LongLLMLingua, a technology designed to compress model text prompts. By utilizing LongLLMLingua, text prompts can be compressed up to 20 times, significantly reducing AI computational costs. Experimental results demonstrate that the compressed prompts not only enhance the model's performance but also decrease latency. This technology holds broad application potential in long-text scenarios.