DLLM-Searcher: Adapting Diffusion Large Language Model for Search Agents

Researchers have developed a new framework called DLLM-Searcher to speed up AI search agents. It uses a special type of AI model and a novel 'Parallel-Reasoning and Acting' method to let the agent think and request information simultaneously, instead of waiting for each step to finish. This approach achieves similar performance to current systems while delivering approximately 15% faster inference, making real-time, complex searches more practical.