Beyond Perfect APIs: A Comprehensive Evaluation of LLM Agents Under Real-World API Complexity

We introduce WildAGTEval, a benchmark designed to evaluate large language model (LLM) agents' function-calling capabilities under realistic API complexity. Unlike prior work that assumes an idealized API system and disregards real-world factors such as noisy API outputs, WildAGTEval accounts for two dimensions of real-world complexity: 1. API specification, which includes detailed documentation and usage constraints, and 2. API execution, which captures runtime challenges. Consequently, WildAGTEval offers (i) an API system encompassing 60 distinct complexity scenarios that can be composed into approximately 32K test configurations, and (ii) user-agent interactions for evaluating LLM agents on these scenarios. Using WildAGTEval, we systematically assess several advanced LLMs and observe that most scenarios are challenging, with irrelevant information complexity posing the greatest difficulty and reducing the performance of strong LLMs by 27.3%. Furthermore, our qualitative analysis reveals that LLMs occasionally distort user intent merely to claim task completion, critically affecting user satisfaction.

MAC: A Multi-Agent Framework for Interactive User Clarification in Multi-turn Conversations

Conversational agents often encounter ambiguous user requests, requiring an effective clarification to successfully complete tasks. While recent advancements in real-world applications favor multi-agent architectures to manage complex conversational scenarios efficiently, ambiguity resolution remains a critical and underexplored challenge--particularly due to the difficulty of determining which agent should initiate a clarification and how agents should coordinate their actions when faced with uncertain or incomplete user input. The fundamental questions of when to interrupt a user and how to formulate the optimal clarification query within the most optimal multi-agent settings remain open. In this paper, we propose MAC (Multi-Agent Clarification), an interactive multi-agent framework specifically optimized to resolve user ambiguities by strategically managing clarification dialogues. We first introduce a novel taxonomy categorizing user ambiguities to systematically guide clarification strategies. Then, we present MAC that autonomously coordinates multiple agents to interact synergistically with users. Empirical evaluations on MultiWOZ 2.4 demonstrate that enabling clarification at both levels increases task success rate 7.8\% (54.5 to 62.3) and reduces the average number of dialogue turns (6.53 to 4.86) by eliciting all required user information up front and minimizing repetition. Our findings highlight the importance of active user interaction and role-aware clarification for more reliable human-agent communication.

SpeakRL: Synergizing Reasoning, Speaking, and Acting in Language Models with Reinforcement Learning

Effective human-agent collaboration is increasingly prevalent in real-world applications. Current trends in such collaborations are predominantly unidirectional, with users providing instructions or posing questions to agents, where agents respond directly without seeking necessary clarifications or confirmations. However, the evolving capabilities of these agents require more proactive engagement, where agents should dynamically participate in conversations to clarify user intents, resolve ambiguities, and adapt to changing circumstances. Existing prior work under-utilize the conversational capabilities of language models (LMs), thereby optimizing agents as better followers rather than effective speakers. In this work, we introduce SpeakRL, a reinforcement learning (RL) method that enhances agents' conversational capabilities by rewarding proactive interactions with users, such as asking right clarification questions when necessary. To support this, we curate SpeakER, a synthetic dataset that includes diverse scenarios from task-oriented dialogues, where tasks are resolved through interactive clarification questions. We present a systematic analysis of reward design for conversational proactivity and propose a principled reward formulation for teaching agents to balance asking with acting. Empirical evaluations demonstrate that our approach achieves a 20.14% absolute improvement in task completion over base models without increasing conversation turns even surpassing even much larger proprietary models, demonstrating the promise of clarification-centric user-agent interactions.

Multimodal Policy Internalization for Conversational Agents

Modern conversational agents like ChatGPT and Alexa+ rely on predefined policies specifying metadata, response styles, and tool-usage rules. As these LLM-based systems expand to support diverse business and user queries, such policies, often implemented as in-context prompts, are becoming increasingly complex and lengthy, making faithful adherence difficult and imposing large fixed computational costs. With the rise of multimodal agents, policies that govern visual and multimodal behaviors are critical but remain understudied. Prior prompt-compression work mainly shortens task templates and demonstrations, while existing policy-alignment studies focus only on text-based safety rules. We introduce Multimodal Policy Internalization (MPI), a new task that internalizes reasoning-intensive multimodal policies into model parameters, enabling stronger policy-following without including the policy during inference. MPI poses unique data and algorithmic challenges. We build two datasets spanning synthetic and real-world decision-making and tool-using tasks and propose TriMPI, a three-stage training framework. TriMPI first injects policy knowledge via continual pretraining, then performs supervised finetuning, and finally applies PolicyRollout, a GRPO-style reinforcement learning extension that augments rollouts with policy-aware responses for grounded exploration. TriMPI achieves notable gains in end-to-end accuracy, generalization, and robustness to forgetting. As the first work on multimodal policy internalization, we provide datasets, training recipes, and comprehensive evaluations to foster future research. Project page: https://mikewangwzhl.github.io/TriMPI.

GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models

We present GLM-4.5, an open-source Mixture-of-Experts (MoE) large language model with 355B total parameters and 32B activated parameters, featuring a hybrid reasoning method that supports both thinking and direct response modes. Through multi-stage training on 23T tokens and comprehensive post-training with expert model iteration and reinforcement learning, GLM-4.5 achieves strong performance across agentic, reasoning, and coding (ARC) tasks, scoring 70.1% on TAU-Bench, 91.0% on AIME 24, and 64.2% on SWE-bench Verified. With much fewer parameters than several competitors, GLM-4.5 ranks 3rd overall among all evaluated models and 2nd on agentic benchmarks. We release both GLM-4.5 (355B parameters) and a compact version, GLM-4.5-Air (106B parameters), to advance research in reasoning and agentic AI systems. Code, models, and more information are available at https://github.com/zai-org/GLM-4.5.

Memetic Search for Green Vehicle Routing Problem with Private Capacitated Refueling Stations

The green vehicle routing problem with private capacitated alternative fuel stations (GVRP-PCAFS) extends the traditional green vehicle routing problem by considering refueling stations limited capacity, where a limited number of vehicles can refuel simultaneously with additional vehicles must wait. This feature presents new challenges for route planning, as waiting times at stations must be managed while keeping route durations within limits and reducing total travel distance. This article presents METS, a novel memetic algorithm (MA) with separate constraint-based tour segmentation (SCTS) and efficient local search (ELS) for solving GVRP-PCAFS. METS combines global and local search effectively through three novelties. For global search, the SCTS strategy splits giant tours to generate diverse solutions, and the search process is guided by a comprehensive fitness evaluation function to dynamically control feasibility and diversity to produce solutions that are both diverse and near-feasible. For local search, ELS incorporates tailored move operators with constant-time move evaluation mechanisms, enabling efficient exploration of large solution neighborhoods. Experimental results demonstrate that METS discovers 31 new best-known solutions out of 40 instances in existing benchmark sets, achieving substantial improvements over current state-of-the-art methods. Additionally, a new large-scale benchmark set based on real-world logistics data is introduced to facilitate future research.

LLM-Virus: Evolutionary Jailbreak Attack on Large Language Models

While safety-aligned large language models (LLMs) are increasingly used as the cornerstone for powerful systems such as multi-agent frameworks to solve complex real-world problems, they still suffer from potential adversarial queries, such as jailbreak attacks, which attempt to induce harmful content. Researching attack methods allows us to better understand the limitations of LLM and make trade-offs between helpfulness and safety. However, existing jailbreak attacks are primarily based on opaque optimization techniques (e.g. token-level gradient descent) and heuristic search methods like LLM refinement, which fall short in terms of transparency, transferability, and computational cost. In light of these limitations, we draw inspiration from the evolution and infection processes of biological viruses and propose LLM-Virus, a jailbreak attack method based on evolutionary algorithm, termed evolutionary jailbreak. LLM-Virus treats jailbreak attacks as both an evolutionary and transfer learning problem, utilizing LLMs as heuristic evolutionary operators to ensure high attack efficiency, transferability, and low time cost. Our experimental results on multiple safety benchmarks show that LLM-Virus achieves competitive or even superior performance compared to existing attack methods.

Multi-scale Temporal Fusion Transformer for Incomplete Vehicle Trajectory Prediction

Motion prediction plays an essential role in autonomous driving systems, enabling autonomous vehicles to achieve more accurate local-path planning and driving decisions based on predictions of the surrounding vehicles. However, existing methods neglect the potential missing values caused by object occlusion, perception failures, etc., which inevitably degrades the trajectory prediction performance in real traffic scenarios. To address this limitation, we propose a novel end-to-end framework for incomplete vehicle trajectory prediction, named Multi-scale Temporal Fusion Transformer (MTFT), which consists of the Multi-scale Attention Head (MAH) and the Continuity Representation-guided Multi-scale Fusion (CRMF) module. Specifically, the MAH leverages the multi-head attention mechanism to parallelly capture multi-scale motion representation of trajectory from different temporal granularities, thus mitigating the adverse effect of missing values on prediction. Furthermore, the multi-scale motion representation is input into the CRMF module for multi-scale fusion to obtain the robust temporal feature of the vehicle. During the fusion process, the continuity representation of vehicle motion is first extracted across time steps to guide the fusion, ensuring that the resulting temporal feature incorporates both detailed information and the overall trend of vehicle motion, which facilitates the accurate decoding of future trajectory that is consistent with the vehicle's motion trend. We evaluate the proposed model on four datasets derived from highway and urban traffic scenarios. The experimental results demonstrate its superior performance in the incomplete vehicle trajectory prediction task compared with state-of-the-art models, e.g., a comprehensive performance improvement of more than 39% on the HighD dataset.

RecMind: Large Language Model Powered Agent For Recommendation

Recent advancements in instructing Large Language Models (LLMs) to utilize external tools and execute multi-step plans have significantly enhanced their ability to solve intricate tasks, ranging from mathematical problems to creative writing. Yet, there remains a notable gap in studying the capacity of LLMs in responding to personalized queries such as a recommendation request. To bridge this gap, we have designed an LLM-powered autonomous recommender agent, RecMind, which is capable of providing precise personalized recommendations through careful planning, utilizing tools for obtaining external knowledge, and leveraging individual data. We propose a novel algorithm, Self-Inspiring, to improve the planning ability of the LLM agent. At each intermediate planning step, the LLM 'self-inspires' to consider all previously explored states to plan for next step. This mechanism greatly improves the model's ability to comprehend and utilize historical planning information for recommendation. We evaluate RecMind's performance in various recommendation scenarios, including rating prediction, sequential recommendation, direct recommendation, explanation generation, and review summarization. Our experiment shows that RecMind outperforms existing zero/few-shot LLM-based recommendation methods in different recommendation tasks and achieves competitive performance to a recent model P5, which requires fully pre-train for the recommendation tasks.

Multi-Task Knowledge Enhancement for Zero-Shot and Multi-Domain Recommendation in an AI Assistant Application

Recommender systems have found significant commercial success but still struggle with integrating new users. Since users often interact with content in different domains, it is possible to leverage a user's interactions in previous domains to improve that user's recommendations in a new one (multi-domain recommendation). A separate research thread on knowledge graph enhancement uses external knowledge graphs to improve single domain recommendations (knowledge graph enhancement). Both research threads incorporate related information to improve predictions in a new domain. We propose in this work to unify these approaches: Using information from interactions in other domains as well as external knowledge graphs to make predictions in a new domain that would be impossible with either information source alone. We apply these ideas to a dataset derived from millions of users' requests for content across three domains (videos, music, and books) in a live virtual assistant application. We demonstrate the advantage of combining knowledge graph enhancement with previous multi-domain recommendation techniques to provide better overall recommendations as well as for better recommendations on new users of a domain.