Generative Engine Marketing (GEM) is an emerging ecosystem for monetizing generative engines, such as LLM-based chatbots, by seamlessly integrating relevant advertisements into their responses. At the core of GEM lies the generation and evaluation of ad-injected responses. However, existing benchmarks are not specifically designed for this purpose, which limits future research. To address this gap, we propose GEM-Bench, the first comprehensive benchmark for ad-injected response generation in GEM. GEM-Bench includes three curated datasets covering both chatbot and search scenarios, a metric ontology that captures multiple dimensions of user satisfaction and engagement, and several baseline solutions implemented within an extensible multi-agent framework. Our preliminary results indicate that, while simple prompt-based methods achieve reasonable engagement such as click-through rate, they often reduce user satisfaction. In contrast, approaches that insert ads based on pre-generated ad-free responses help mitigate this issue but introduce additional overhead. These findings highlight the need for future research on designing more effective and efficient solutions for generating ad-injected responses in GEM.
Peer review remains the central quality-control mechanism of science, yet its ability to fulfill this role is increasingly strained. Empirical studies document serious shortcomings: long publication delays, escalating reviewer burden concentrated on a small minority of scholars, inconsistent quality and low inter-reviewer agreement, and systematic biases by gender, language, and institutional prestige. Decades of human-centered reforms have yielded only marginal improvements. Meanwhile, artificial intelligence, especially large language models (LLMs), is being piloted across the peer-review pipeline by journals, funders, and individual reviewers. Early studies suggest that AI assistance can produce reviews comparable in quality to humans, accelerate reviewer selection and feedback, and reduce certain biases, but also raise distinctive concerns about hallucination, confidentiality, gaming, novelty recognition, and loss of trust. In this paper, we map the aims and persistent failure modes of peer review to specific LLM applications and systematically analyze the objections they raise alongside safeguards that could make their use acceptable. Drawing on emerging evidence, we show that targeted, supervised LLM assistance can plausibly improve error detection, timeliness, and reviewer workload without displacing human judgment. We highlight advanced architectures, including fine-tuned, retrieval-augmented, and multi-agent systems, that may enable more reliable, auditable, and interdisciplinary review. We argue that ethical and practical considerations are not peripheral but constitutive: the legitimacy of AI-assisted peer review depends on governance choices as much as technical capacity. The path forward is neither uncritical adoption nor reflexive rejection, but carefully scoped pilots with explicit evaluation metrics, transparency, and accountability.
Personalized financial advice requires consideration of user goals, constraints, risk tolerance, and jurisdiction. Prior LLM work has focused on support systems for investors and financial planners. Simultaneously, numerous recent studies examine broader personal finance tasks, including budgeting, debt management, retirement, and estate planning, through agentic pipelines that incur high maintenance costs, yielding less than 25% of their expected financial returns. In this study, we introduce a novel and reproducible framework that integrates relevant financial context with behavioral finance studies to construct supervision data for end-to-end advisors. Using this framework, we create a 19k sample reasoning dataset and conduct a comprehensive fine-tuning of the Qwen-3-8B model on the dataset. Through a held-out test split and a blind LLM-jury study, we demonstrate that through careful data curation and behavioral integration, our 8B model achieves performance comparable to significantly larger baselines (14-32B parameters) across factual accuracy, fluency, and personalization metrics while incurring 80% lower costs than the larger counterparts.
Analog and mixed-signal circuit design remains challenging due to the shortage of high-quality data and the difficulty of embedding domain knowledge into automated flows. Traditional black-box optimization achieves sampling efficiency but lacks circuit understanding, which often causes evaluations to be wasted in low-value regions of the design space. In contrast, learning-based methods embed structural knowledge but are case-specific and costly to retrain. Recent attempts with large language models show potential, yet they often rely on manual intervention, limiting generality and transparency. We propose TopoSizing, an end-to-end framework that performs robust circuit understanding directly from raw netlists and translates this knowledge into optimization gains. Our approach first applies graph algorithms to organize circuits into a hierarchical device-module-stage representation. LLM agents then execute an iterative hypothesis-verification-refinement loop with built-in consistency checks, producing explicit annotations. Verified insights are integrated into Bayesian optimization through LLM-guided initial sampling and stagnation-triggered trust-region updates, improving efficiency while preserving feasibility.
While virtual reality (VR) excels at simulating physical environments, its effectiveness for training complex interpersonal skills is limited by a lack of psychologically plausible virtual humans. This is a critical gap in high-stakes domains like medical education, where communication is a core competency. This paper introduces a framework that integrates large language models (LLMs) into immersive VR to create medically coherent virtual patients with distinct, consistent personalities, built on a modular architecture that decouples personality from clinical data. We evaluated our system in a mixed-method, within-subjects study with licensed physicians who engaged in simulated consultations. Results demonstrate that the approach is not only feasible but is also perceived by physicians as a highly rewarding and effective training enhancement. Furthermore, our analysis uncovers critical design principles, including a ``realism-verbosity paradox" where less communicative agents can seem more artificial, and the need for challenges to be perceived as authentic to be instructive. This work provides a validated framework and key insights for developing the next generation of socially intelligent VR training environments.
Generative Large Language Models (LLMs) have demonstrated remarkable performance across a wide range of tasks. Recent research has introduced Multi-Agent Debate (MAD) systems, which leverage multiple LLMs to simulate human debate and thereby improve task performance. However, while some LLMs may possess superior knowledge or reasoning capabilities for specific tasks, they often struggle to clearly communicate this advantage during debates, in part due to a lack of confidence expression. Moreover, inappropriate confidence expression can cause agents in MAD systems to either stubbornly maintain incorrect beliefs or converge prematurely on suboptimal answers, ultimately reducing debate effectiveness and overall system performance. To address these challenges, we propose incorporating confidence expression into MAD systems to allow LLMs to explicitly communicate their confidence levels. To validate this approach, we develop ConfMAD, a MAD framework that integrates confidence expression throughout the debate process. Experimental results demonstrate the effectiveness of our method, and we further analyze how confidence influences debate dynamics, offering insights into the design of confidence-aware MAD systems.
High-quality annotated data is a cornerstone of modern Natural Language Processing (NLP). While recent methods begin to leverage diverse annotation sources-including Large Language Models (LLMs), Small Language Models (SLMs), and human experts-they often focus narrowly on the labeling step itself. A critical gap remains in the holistic process control required to manage these sources dynamically, addressing complex scheduling and quality-cost trade-offs in a unified manner. Inspired by real-world crowdsourcing companies, we introduce CrowdAgent, a multi-agent system that provides end-to-end process control by integrating task assignment, data annotation, and quality/cost management. It implements a novel methodology that rationally assigns tasks, enabling LLMs, SLMs, and human experts to advance synergistically in a collaborative annotation workflow. We demonstrate the effectiveness of CrowdAgent through extensive experiments on six diverse multimodal classification tasks. The source code and video demo are available at https://github.com/QMMMS/CrowdAgent.
Modern scientific discovery increasingly relies on workflows that process data across the Edge, Cloud, and High Performance Computing (HPC) continuum. Comprehensive and in-depth analyses of these data are critical for hypothesis validation, anomaly detection, reproducibility, and impactful findings. Although workflow provenance techniques support such analyses, at large scale, the provenance data become complex and difficult to analyze. Existing systems depend on custom scripts, structured queries, or static dashboards, limiting data interaction. In this work, we introduce an evaluation methodology, reference architecture, and open-source implementation that leverages interactive Large Language Model (LLM) agents for runtime data analysis. Our approach uses a lightweight, metadata-driven design that translates natural language into structured provenance queries. Evaluations across LLaMA, GPT, Gemini, and Claude, covering diverse query classes and a real-world chemistry workflow, show that modular design, prompt tuning, and Retrieval-Augmented Generation (RAG) enable accurate and insightful LLM agent responses beyond recorded provenance.
[Background] Large Language Model (LLM)-based multi-agent systems (MAS) are transforming software development by enabling autonomous collaboration. Classical software processes such asWaterfall, V-Model, and Agile offer structured coordination patterns that can be repurposed to guide these agent interactions. [Aims] This study explores how traditional software development processes can be adapted as coordination scaffolds for LLM based MAS and examines their impact on code quality, cost, and productivity. [Method] We executed 11 diverse software projects under three process models and four GPT variants, totaling 132 runs. Each output was evaluated using standardized metrics for size (files, LOC), cost (execution time, token usage), and quality (code smells, AI- and human detected bugs). [Results] Both process model and LLM choice significantly affected system performance. Waterfall was most efficient, V-Model produced the most verbose code, and Agile achieved the highest code quality, albeit at higher computational cost. [Conclusions] Classical software processes can be effectively instantiated in LLM-based MAS, but each entails trade-offs across quality, cost, and adaptability. Process selection should reflect project goals, whether prioritizing efficiency, robustness, or structured validation.
Automated issue solving seeks to autonomously identify and repair defective code snippets across an entire codebase. SWE-Bench has emerged as the most widely adopted benchmark for evaluating progress in this area. While LLM-based agentic tools show great promise, they still fail on a substantial portion of tasks. Moreover, current evaluations primarily report aggregate issue-solving rates, which obscure the underlying causes of success and failure, making it challenging to diagnose model weaknesses or guide targeted improvements. To bridge this gap, we first analyze the performance and efficiency of three SOTA tools, spanning both pipeline-based and agentic architectures, in automated issue solving tasks of SWE-Bench-Verified under varying task characteristics. Furthermore, to move from high-level performance metrics to underlying cause analysis, we conducted a systematic manual analysis of 150 failed instances. From this analysis, we developed a comprehensive taxonomy of failure modes comprising 3 primary phases, 9 main categories, and 25 fine-grained subcategories. Then we systematically analyze the distribution of the identified failure modes, the results reveal distinct failure fingerprints between the two architectural paradigms, with the majority of agentic failures stemming from flawed reasoning and cognitive deadlocks. Motivated by these insights, we propose a collaborative Expert-Executor framework. It introduces a supervisory Expert agent tasked with providing strategic oversight and course-correction for a primary Executor agent. This architecture is designed to correct flawed reasoning and break the cognitive deadlocks that frequently lead to failure. Experiments show that our framework solves 22.2% of previously intractable issues for a leading single agent. These findings pave the way for building more robust agents through diagnostic evaluation and collaborative design.
Large language models (LLMs) can lead to undesired consequences when misaligned with human values, especially in scenarios involving complex and sensitive social biases. Previous studies have revealed the misalignment of LLMs with human values using expert-designed or agent-based emulated bias scenarios. However, it remains unclear whether the alignment of LLMs with human values differs across different types of scenarios (e.g., scenarios containing negative vs. non-negative questions). In this study, we investigate the alignment of LLMs with human values regarding social biases (HVSB) in different types of bias scenarios. Through extensive analysis of 12 LLMs from four model families and four datasets, we demonstrate that LLMs with large model parameter scales do not necessarily have lower misalignment rate and attack success rate. Moreover, LLMs show a certain degree of alignment preference for specific types of scenarios and the LLMs from the same model family tend to have higher judgment consistency. In addition, we study the understanding capacity of LLMs with their explanations of HVSB. We find no significant differences in the understanding of HVSB across LLMs. We also find LLMs prefer their own generated explanations. Additionally, we endow smaller language models (LMs) with the ability to explain HVSB. The generation results show that the explanations generated by the fine-tuned smaller LMs are more readable, but have a relatively lower model agreeability.
Large Language Models (LLMs) have made remarkable progress in mathematical reasoning, but still continue to struggle with high-precision tasks like numerical computation and formal symbolic manipulation. Integrating external tools has emerged as a promising approach to bridge this gap. Despite recent advances, existing methods struggle with three key challenges: constructing tool-integrated reasoning data, performing fine-grained optimization, and enhancing inference. To overcome these limitations, we propose THOR (Tool-Integrated Hierarchical Optimization via RL). First, we introduce TIRGen, a multi-agent actor-critic-based pipeline for constructing high-quality datasets of tool-integrated reasoning paths, aligning with the policy and generalizing well across diverse models. Second, to perform fine-grained hierarchical optimization, we introduce an RL strategy that jointly optimizes for both trajectory-level problem solving and step-level code generation. This is motivated by our key insight that the success of an intermediate tool call is a strong predictor of the final answer's correctness. Finally, THOR incorporates a self-correction mechanism that leverages immediate tool feedback to dynamically revise erroneous reasoning paths during inference. Our approach demonstrates strong generalization across diverse models, performing effectively in both reasoning and non-reasoning models. It further achieves state-of-the-art performance for models of a similar scale on multiple mathematical benchmarks, while also delivering consistent improvements on code benchmarks. Our code will be publicly available at https://github.com/JingMog/THOR.
LLM-based multi-agent simulations are a rapidly growing field of research, but current simulations often lack clear modes for interaction and analysis, limiting the "what if" scenarios researchers are able to investigate. In this demo, we define three core operations for interacting with multi-agent simulations: inject, fork, and compare. Inject allows researchers to introduce external events at any point during simulation execution. Fork creates independent timeline branches from any timestamp, preserving complete state while allowing divergent exploration. Compare facilitates parallel observation of multiple branches, revealing how different interventions lead to distinct emergent behaviors. Together, these operations establish a vocabulary that transforms linear simulation workflows into interactive, explorable spaces. We demonstrate this vocabulary through a commodity market simulation with fourteen AI agents, where researchers can inject contrasting events and observe divergent outcomes across parallel timelines. By defining these fundamental operations, we provide a starting point for systematic causal investigation in LLM-based agent simulations, moving beyond passive observation toward active experimentation.
Mission-critical industrial infrastructure, such as data centers, increasingly depends on complex management software. Its operations, however, pose significant challenges due to the escalating system complexity, multi-vendor integration, and a shortage of expert operators. While Robotic Process Automation (RPA) offers partial automation through handcrafted scripts, it suffers from limited flexibility and high maintenance costs. Recent advances in Large Language Model (LLM)-based graphical user interface (GUI) agents have enabled more flexible automation, yet these general-purpose agents face five critical challenges when applied to industrial management, including unfamiliar element understanding, precision and efficiency, state localization, deployment constraints, and safety requirements. To address these issues, we propose InfraMind, a novel exploration-based GUI agentic framework specifically tailored for industrial management systems. InfraMind integrates five innovative modules to systematically resolve different challenges in industrial management: (1) systematic search-based exploration with virtual machine snapshots for autonomous understanding of complex GUIs; (2) memory-driven planning to ensure high-precision and efficient task execution; (3) advanced state identification for robust localization in hierarchical interfaces; (4) structured knowledge distillation for efficient deployment with lightweight models; and (5) comprehensive, multi-layered safety mechanisms to safeguard sensitive operations. Extensive experiments on both open-source and commercial DCIM platforms demonstrate that our approach consistently outperforms existing frameworks in terms of task success rate and operational efficiency, providing a rigorous and scalable solution for industrial management automation.
We propose LLM-Interleaved (LLM-I), a flexible and dynamic framework that reframes interleaved image-text generation as a tool-use problem. LLM-I is designed to overcome the "one-tool" bottleneck of current unified models, which are limited to synthetic imagery and struggle with tasks requiring factual grounding or programmatic precision. Our framework empowers a central LLM or MLLM agent to intelligently orchestrate a diverse toolkit of specialized visual tools, including online image search, diffusion-based generation, code execution, and image editing. The agent is trained to select and apply these tools proficiently via a Reinforcement Learning (RL) framework that features a hybrid reward system combining rule-based logic with judgments from LLM and MLLM evaluators. Trained on a diverse new dataset using four different model backbones, LLM-I demonstrates state-of-the-art performance, outperforming existing methods by a large margin across four benchmarks. We also introduce a novel test-time scaling strategy that provides further performance gains. Project Page: https://github.com/ByteDance-BandAI/LLM-I.
Autonomous LLM agents can issue thousands of API calls per hour without human oversight. OAuth 2.0 assumes deterministic clients, but in agentic settings stochastic reasoning, prompt injection, or multi-agent orchestration can silently expand privileges. We introduce Agentic JWT (A-JWT), a dual-faceted intent token that binds each agent's action to verifiable user intent and, optionally, to a specific workflow step. A-JWT carries an agent's identity as a one-way checksum hash derived from its prompt, tools and configuration, and a chained delegation assertion to prove which downstream agent may execute a given task, and per-agent proof-of-possession keys to prevent replay and in-process impersonation. We define a new authorization mechanism and add a lightweight client shim library that self-verifies code at run time, mints intent tokens, tracks workflow steps and derives keys, thus enabling secure agent identity and separation even within a single process. We illustrate a comprehensive threat model for agentic applications, implement a Python proof-of-concept and show functional blocking of scope-violating requests, replay, impersonation, and prompt-injection pathways with sub-millisecond overhead on commodity hardware. The design aligns with ongoing OAuth agent discussions and offers a drop-in path toward zero-trust guarantees for agentic applications. A comprehensive performance and security evaluation with experimental results will appear in our forthcoming journal publication
This paper introduces CoBRA, a novel toolkit for systematically specifying agent behavior in LLM-based social simulation. We found that conventional approaches that specify agent behaviors through implicit natural language descriptions cannot yield consistent behaviors across models, and the produced agent behaviors do not capture the nuances of the descriptions. In contrast, CoBRA presents a new approach to program agents' cognitive biases explicitly, by grounding agents' expected behaviors using classic social science experiments. CoBRA has two components: (1) Cognitive Bias Index that measures the cognitive bias of a social agent, by quantifying the agent's reactions in a set of validated classical social science experiments; (2) Behavioral Regulation Engine that aligns the agent's behavior to demonstrate controlled cognitive bias. We evaluated CoBRA as an HCI toolkit through demonstration and technical benchmarks. Our results suggest that CoBRA can precisely program the cognitive bias demonstrated in a social agent in a model-agnostic manner.
Coarse-grained Reconfigurable Arrays (CGRAs) are a promising computing architecture that can deliver high-performance, energy-efficient acceleration across diverse domains. By supporting reconfiguration at the functional unit level, CGRAs efficiently adapt to varying computational patterns and optimize resource utilization. However, designing CGRAs is highly challenging due to the vast design space, independent architectural parameters, and the time-consuming nature of manual design. Fortunately, the rapid advancement of large language models (LLMs) presents new opportunities to automate this process. In this work, we propose MACO -- an open-source multi-agent LLM-based framework for Hardware/Software (HW/SW) co-design of CGRAs. The framework employs LLM reasoning to generate CGRAs across four stages: HW/SW co-design, Design error correction, Best design selection, and Evaluation & Feedback. Furthermore, MACO iteratively optimizes the generated CGRAs, leveraging agent reasoning and feedback to achieve higher PPA (that is, power, performance, and area) design points for a given domain. In addition, we introduce an LLM self-learning mechanism that employs LLM-driven decision making to select the optimal CGRA to accelerate the design process. We evaluate the framework with state-of-the-art LLM-based methods and manual CGRA design, in terms of performance, power consumption, and area. Experimental results show that MACO efficiently generates high-quality CGRA architectures, significantly reducing manual design effort and demonstrating the potential of our framework for real-world CGRA design.
We investigate whether giving LLM agents the collaborative tools and autonomy that humans naturally use for problem solving can improve their performance. We equip Claude Code agents with MCP-based social media and journaling tools and allow them to use these tools as they see fit. Across 34 Aider Polyglot Python programming challenges, collaborative tools substantially improve performance on the hardest problems, delivering 15-40% lower cost, 12-27% fewer turns, and 12-38% faster completion than baseline agents. Effects on the full challenge set are mixed, suggesting these tools act as performance enhancers when additional reasoning scaffolding is most needed. Surprisingly, Different models naturally adopted distinct collaborative strategies without explicit instruction. Sonnet 3.7 engaged broadly across tools and benefited from articulation-based cognitive scaffolding. Sonnet 4 showed selective adoption, leaning on journal-based semantic search when problems were genuinely difficult. This mirrors how human developers adjust collaboration based on expertise and task complexity. Behavioral analysis shows agents prefer writing over reading by about 2-9x, indicating that structured articulation drives much of the improvement rather than information access alone. Overall, AI agents can systematically benefit from human-inspired collaboration tools at the edge of their capabilities, pointing to adaptive collaborative interfaces as reasoning enhancers rather than universal efficiency boosts.
Crash reports are central to software maintenance, yet many lack the diagnostic detail developers need to debug efficiently. We examine whether large language models can enhance crash reports by adding fault locations, root-cause explanations, and repair suggestions. We study two enhancement strategies: Direct-LLM, a single-shot approach that uses stack-trace context, and Agentic-LLM, an iterative approach that explores the repository for additional evidence. On a dataset of 492 real-world crash reports, LLM-enhanced reports improve Top-1 problem-localization accuracy from 10.6% (original reports) to 40.2-43.1%, and produce suggested fixes that closely resemble developer patches (CodeBLEU around 56-57%). Both our manual evaluations and LLM-as-a-judge assessment show that Agentic-LLM delivers stronger root-cause explanations and more actionable repair guidance. A user study with 16 participants further confirms that enhanced reports make crashes easier to understand and resolve, with the largest improvement in repair guidance. These results indicate that supplying LLMs with stack traces and repository code yields enhanced crash reports that are substantially more useful for debugging.
Large language models (LLMs) show promise for translating natural-language statutes into executable logic, but reliability in legally critical settings remains challenging due to ambiguity and hallucinations. We present an agentic approach for developing legal-critical software, using U.S. federal tax preparation as a case study. The key challenge is test-case generation under the oracle problem, where correct outputs require interpreting law. Building on metamorphic testing, we introduce higher-order metamorphic relations that compare system outputs across structured shifts among similar individuals. Because authoring such relations is tedious and error-prone, we use an LLM-driven, role-based framework to automate test generation and code synthesis. We implement a multi-agent system that translates tax code into executable software and incorporates a metamorphic-testing agent that searches for counterexamples. In experiments, our framework using a smaller model (GPT-4o-mini) achieves a worst-case pass rate of 45%, outperforming frontier models (GPT-4o and Claude 3.5, 9-15%) on complex tax-code tasks. These results support agentic LLM methodologies as a path to robust, trustworthy legal-critical software from natural-language specifications.
Large Language Models are reshaping task automation, yet remain limited in complex, multi-step real-world tasks that require aligning with vague user intent and enabling dynamic user override. From a formative study with 12 participants, we found that end-users actively seek to shape generative interfaces rather than relying on one-shot outputs. To address this, we introduce the human-agent co-generation paradigm, materialized in DuetUI. This LLM-empowered system unfolds alongside task progress through a bidirectional context loop--the agent scaffolds the interface by decomposing the task, while the user's direct manipulations implicitly steer the agent's next generation step. In a user study with 24 participants, DuetUI significantly improved task efficiency and interface usability compared to a baseline, fostering seamless human-agent collaboration. Our contributions include the proposal and validation of this novel paradigm, the design of the DuetUI prototype embodying it, and empirical insights into how this bidirectional loop better aligns agents with human intent.
Large Language Model (LLM)-based web agents demonstrate strong performance on knowledge-intensive tasks but are hindered by context window limitations in paradigms like ReAct. Complex queries involving multiple entities, intertwined relationships, and high uncertainty demand extensive search cycles that rapidly exhaust context budgets before reaching complete solutions. To overcome this challenge, we introduce ReSum, a novel paradigm that enables indefinite exploration through periodic context summarization. ReSum converts growing interaction histories into compact reasoning states, maintaining awareness of prior discoveries while bypassing context constraints. For paradigm adaptation, we propose ReSum-GRPO, integrating GRPO with segmented trajectory training and advantage broadcasting to familiarize agents with summary-conditioned reasoning. Extensive experiments on web agents of varying scales across three benchmarks demonstrate that ReSum delivers an average absolute improvement of 4.5\% over ReAct, with further gains of up to 8.2\% following ReSum-GRPO training. Notably, with only 1K training samples, our WebResummer-30B (a ReSum-GRPO-trained version of WebSailor-30B) achieves 33.3\% Pass@1 on BrowseComp-zh and 18.3\% on BrowseComp-en, surpassing existing open-source web agents.
Large language models (LLMs) have evolved into agentic systems capable of autonomous tool use and multi-step reasoning for complex problem-solving. However, post-training approaches building upon general-purpose foundation models consistently underperform in agentic tasks, particularly in open-source implementations. We identify the root cause: the absence of robust agentic foundation models forces models during post-training to simultaneously learn diverse agentic behaviors while aligning them to expert demonstrations, thereby creating fundamental optimization tensions. To this end, we are the first to propose incorporating Agentic Continual Pre-training (Agentic CPT) into the deep research agents training pipeline to build powerful agentic foundational models. Based on this approach, we develop a deep research agent model named AgentFounder. We evaluate our AgentFounder-30B on 10 benchmarks and achieve state-of-the-art performance while retains strong tool-use ability, notably 39.9% on BrowseComp-en, 43.3% on BrowseComp-zh, and 31.5% Pass@1 on HLE.
Transcending human cognitive limitations represents a critical frontier in LLM training. Proprietary agentic systems like DeepResearch have demonstrated superhuman capabilities on extremely complex information-seeking benchmarks such as BrowseComp, a feat previously unattainable. We posit that their success hinges on a sophisticated reasoning pattern absent in open-source models: the ability to systematically reduce extreme uncertainty when navigating vast information landscapes. Based on this insight, we introduce WebSailor, a complete post-training methodology designed to instill this crucial capability. Our approach involves generating novel, high-uncertainty tasks through structured sampling and information obfuscation, RFT cold start, and an efficient agentic RL training algorithm, Duplicating Sampling Policy Optimization (DUPO). With this integrated pipeline, WebSailor significantly outperforms all open-source agents in complex information-seeking tasks, matching proprietary agents' performance and closing the capability gap.
Large Language Models (LLMs) are increasingly deployed in roles requiring nuanced psychological understanding, such as emotional support agents, counselors, and decision-making assistants. However, their ability to interpret human personality traits, a critical aspect of such applications, remains unexplored, particularly in ecologically valid conversational settings. While prior work has simulated LLM "personas" using discrete Big Five labels on social media data, the alignment of LLMs with continuous, ground-truth personality assessments derived from natural interactions is largely unexamined. To address this gap, we introduce a novel benchmark comprising semi-structured interview transcripts paired with validated continuous Big Five trait scores. Using this dataset, we systematically evaluate LLM performance across three paradigms: (1) zero-shot and chain-of-thought prompting with GPT-4.1 Mini, (2) LoRA-based fine-tuning applied to both RoBERTa and Meta-LLaMA architectures, and (3) regression using static embeddings from pretrained BERT and OpenAI's text-embedding-3-small. Our results reveal that all Pearson correlations between model predictions and ground-truth personality traits remain below 0.26, highlighting the limited alignment of current LLMs with validated psychological constructs. Chain-of-thought prompting offers minimal gains over zero-shot, suggesting that personality inference relies more on latent semantic representation than explicit reasoning. These findings underscore the challenges of aligning LLMs with complex human attributes and motivate future work on trait-specific prompting, context-aware modeling, and alignment-oriented fine-tuning.
Search has emerged as core infrastructure for LLM-based agents and is widely viewed as critical on the path toward more general intelligence. Finance is a particularly demanding proving ground: analysts routinely conduct complex, multi-step searches over time-sensitive, domain-specific data, making it ideal for assessing both search proficiency and knowledge-grounded reasoning. Yet no existing open financial datasets evaluate data searching capability of end-to-end agents, largely because constructing realistic, complicated tasks requires deep financial expertise and time-sensitive data is hard to evaluate. We present FinSearchComp, the first fully open-source agent benchmark for realistic, open-domain financial search and reasoning. FinSearchComp comprises three tasks -- Time-Sensitive Data Fetching, Simple Historical Lookup, and Complex Historical Investigation -- closely reproduce real-world financial analyst workflows. To ensure difficulty and reliability, we engage 70 professional financial experts for annotation and implement a rigorous multi-stage quality-assurance pipeline. The benchmark includes 635 questions spanning global and Greater China markets, and we evaluate 21 models (products) on it. Grok 4 (web) tops the global subset, approaching expert-level accuracy. DouBao (web) leads on the Greater China subset. Experimental analyses show that equipping agents with web search and financial plugins substantially improves results on FinSearchComp, and the country origin of models and tools impact performance significantly.By aligning with realistic analyst tasks and providing end-to-end evaluation, FinSearchComp offers a professional, high-difficulty testbed for complex financial search and reasoning.
Recent advancements in Large Language Models(LLMs) have led to the development of LLM-based AI agents. A key challenge is the creation of agents that can effectively ground themselves in complex, adversarial long-horizon environments. Existing methods mainly focus on (1) using LLMs as policies to interact with the environment through generating low-level feasible actions, and (2) utilizing LLMs to generate high-level tasks or language guides to stimulate action generation. However, the former struggles to generate reliable actions, while the latter relies heavily on expert experience to translate high-level tasks into specific action sequences. To address these challenges, we introduce the Plan with Language, Act with Parameter (PLAP) planning framework that facilitates the grounding of LLM-based agents in long-horizon environments. The PLAP method comprises three key components: (1) a skill library containing environment-specific parameterized skills, (2) a skill planner powered by LLMs, and (3) a skill executor converting the parameterized skills into executable action sequences. We implement PLAP in MicroRTS, a long-horizon real-time strategy game that provides an unfamiliar and challenging environment for LLMs. The experimental results demonstrate the effectiveness of PLAP. In particular, GPT-4o-driven PLAP in a zero-shot setting outperforms 80% of baseline agents, and Qwen2-72B-driven PLAP, with carefully crafted few-shot examples, surpasses the top-tier scripted agent, CoacAI. Additionally, we design comprehensive evaluation metrics and test 6 closed-source and 2 open-source LLMs within the PLAP framework, ultimately releasing an LLM leaderboard ranking long-horizon skill planning ability. Our code is available at https://github.com/AI-Research-TeamX/PLAP.
This work introduces xOffense, an AI-driven, multi-agent penetration testing framework that shifts the process from labor-intensive, expert-driven manual efforts to fully automated, machine-executable workflows capable of scaling seamlessly with computational infrastructure. At its core, xOffense leverages a fine-tuned, mid-scale open-source LLM (Qwen3-32B) to drive reasoning and decision-making in penetration testing. The framework assigns specialized agents to reconnaissance, vulnerability scanning, and exploitation, with an orchestration layer ensuring seamless coordination across phases. Fine-tuning on Chain-of-Thought penetration testing data further enables the model to generate precise tool commands and perform consistent multi-step reasoning. We evaluate xOffense on two rigorous benchmarks: AutoPenBench and AI-Pentest-Benchmark. The results demonstrate that xOffense consistently outperforms contemporary methods, achieving a sub-task completion rate of 79.17%, decisively surpassing leading systems such as VulnBot and PentestGPT. These findings highlight the potential of domain-adapted mid-scale LLMs, when embedded within structured multi-agent orchestration, to deliver superior, cost-efficient, and reproducible solutions for autonomous penetration testing.
Large Language Models (LLMs) have revolutionized the simulation of agent societies, enabling autonomous planning, memory formation, and social interactions. However, existing frameworks often overlook systematic evaluations for event organization and lack visualized integration with physically grounded environments, limiting agents' ability to navigate spaces and interact with items realistically. We develop MiniAgentPro, a visualization platform featuring an intuitive map editor for customizing environments and a simulation player with smooth animations. Based on this tool, we introduce a comprehensive test set comprising eight diverse event scenarios with basic and hard variants to assess agents' ability. Evaluations using GPT-4o demonstrate strong performance in basic settings but highlight coordination challenges in hard variants.