The field of deep research is rapidly evolving, with a focus on developing systems that can automate complex, open-ended tasks. Recent developments have centered around the integration of large language models (LLMs) with external tools, such as search engines, to enable more effective information retrieval and synthesis. This has led to the creation of deep research agents (DRAs) that can produce analyst-level reports through iterative information retrieval and synthesis. Notably, researchers are exploring new evaluation paradigms and benchmarks to assess the performance of DRAs, highlighting the need for more comprehensive and systematic evaluation frameworks. Some notable papers have made significant contributions to this area, including the introduction of novel benchmarks and evaluation frameworks. For example, Dr.Mi-Bench and FINDER have been proposed as modular-integrated benchmarks for scientific DR agents, while CAIRNS has demonstrated the importance of balancing readability and scientific accuracy in climate adaptation question answering. Additionally, the Static-DRA has shown promise as a configurable and static deep research agent, offering a pragmatic and resource-aware solution for users.