Search Agents (SAs) typically leverage large language models (LLMs) to support complex information-seeking tasks by autonomously exploring web sources and synthesizing information into comprehensive responses. For SAs evaluation, prior benchmarks mainly focus on specialized tasks that are unlikely to arise in real-world user scenarios. Moreover, their reliance on coarse task-level rubrics often limits evaluation interpretability. To bridge this gap, we introduce DailyReport, an open-ended benchmark to evaluate SA capabilities on daily search tasks. It contains 150 open-ended tasks with 3,546 associated rubrics, capturing widely discussed and timely information demands of real-world users. Each task is decomposed into subtasks and evaluated with cascade rubrics across disentangled dimensions. Through cascade performance attribution and user-centric aggregation, we derive highly interpretable scores for each dimension, along with a user preference score. Our results on 17 agentic systems show that current systems still fall short of users’ expectations. DailyReport will be open-sourced to facilitate future research.

Multimodal large language models (MLLMs) are frequently exposed to auxiliary textual context, the impact of which on vision-language tasks remains underexplored. In this paper, we investigate the influence of task-irrelevant context by formulating it as a controlled intervention within a binary visual judgment framework. Through controlled paired comparisons with and without auxiliary textual context, we observe that irrelevant text consistently biases model predictions across diverse benchmarks. To move beyond performance metrics, we characterize this sensitivity through "decision margin", which is defined by the log-probability difference between candidates. Our analysis reveals an empirical pattern: context-conditioned margins are often well approximated by an affine transformation of their context-free counterparts. We further interpret the fitted affine parameters as metrics for visual commitment preservation and directional answer bias. These findings provide a margin-level diagnostic view of irrelevant-context effects in MLLMs, indicating that irrelevant context does not manifest as stochastic noise, but as an estimable distortion of model preference.

Retrieval-augmented generation (RAG) effectively mitigates hallucinations in large language models (LLMs) by filling knowledge gaps with retrieved external information. Most existing studies primarily retrieve knowledge documents based on semantic similarity to assist in answering questions but ignore the fine-grained necessary information within documents. In this paper, we propose a novel fine-grained knowledge enhancement method (FKE) for RAG, where fine-grained knowledge primarily includes sentence-level information easily overlooked in the document-based retrieval process. Concretely, we create a disentangled Chain-of-Thought prompting procedure to retrieve fine-grained knowledge from the external knowledge corpus. Then we develop a decoding enhancement strategy to constrain the document-based decoding process using fine-grained knowledge, thereby facilitating more accurate generated answers. Given an existing RAG pipeline, our method could be applied in a plug-and-play manner to enhance its performance with no additional modules or training process. Extensive experiments verify the effectiveness and generality of our method.