Simthesizer introduces a composable simulator infrastructure that uniformly expresses the complete serving workflow, including the control decisions that coordinate it, and realizes it as a unified dynamic graph in Simthesizer simulator.
Abstract
System-level simulation is an essential tool for exploring the rapidly expanding design space of LLM serving systems, where real deployments remain costly and often infeasible. However, modern LLM serving now evolves faster than human-driven simulator development can track, and emerging workloads and mechanisms, from agentic workflows to disaggregated serving, no longer fit the monolithic simulation pipeline that existing simulators assume. Each new mechanism therefore demands an invasive rewrite, leaving a widening development gap between deployed serving systems and the simulators that model them. To close this gap, we present Simthesizer, a framework that realizes agent-driven simulator development. Simthesizer introduces a composable simulator infrastructure that uniformly expresses the complete serving workflow, including the control decisions that coordinate it, and realizes it as a unified dynamic graph in Simthesizer simulator. Synthesizer agent, a harnessed coding agent, then lowers natural-language feature requests onto this abstraction under simulator-specific guardrails and fidelity validation, evolving one shared simulator instead of building a new one for every feature. Under the same coding agent and harnesses, extensions built on Simthesizer follow a vLLM-based real system with 2.51% average throughput error, versus 6.03% for extensions built on existing simulators. On identical workloads, Simthesizer also simulates up to 284.96x and 23.19x faster than two state-of-the-art simulators, LLMServingSim2.0 and Vidur, respectively.
This work presents Aries, a full-stack experimentation framework that separates task semantics from execution configurations, reconstructs cross-component agent trajectories with correlated system telemetry, and exposes stateful tool execution through a consistent interface across heterogeneous sandbox substrates.
Leonid Kondrashov, Hongrui Liu, JooYoung Park et al.· 1 citation
Integrated sensing and communications (ISAC) is moving from proof-of-concept demonstrations to system-level deployment in sixth-generation (6G) networks. Because sensing and communication share hardware, spectrum, and waveform resources, ISAC design now involves many tightly coupled choices, including waveform selection, sensing algorithm setup, resource scheduling, and deployment planning. This design space is already too large to manage well through manual tuning or isolated optimizers. This article introduces the \textit{Agent Compiler}, a large language model (LLM)-enabled compilation layer that translates high-level engineering intent into complete and executable ISAC system configurations. The Agent Compiler works in four stages: intent parsing, task decomposition, policy graph synthesis, and infrastructure mapping. It produces a verifiable intermediate representation called the ISAC Policy Graph (IPG). A runtime engine then deploys the compiled configuration and supports closed-loop adaptation at three levels: fast parameter updates, partial recompilation of affected subgraphs, and full workflow recompilation. The core design principle is strict time-scale separation: the LLM handles slow-loop strategic decisions, while proven algorithms retain real-time control in the fast loop. A UAV-assisted disaster rescue example illustrates the full compilation process. We also discuss open issues, including compilation latency, output reliability, constraint verification, and pipeline security, to guide future research.
Lijie Zheng, Xudong Zhong, Baoquan Ren et al.· 0 citations
This work introduces Model Automated Deployment Engine (MADE), a dual-agent coordination system that iteratively constructs and validates the deployment artifacts, updates its deployment belief based on execution feedback, and revisits invalid upstream artifacts until the model is successfully served as a ready-to-call API that can then be used by other agents.
Yicheng Liu, Bolin Zhang, Weiran Liu et al.· 0 citations
Agent-based models (ABMs) have long been employed to explore how individual behaviors aggregate into complex societal phenomena in urban space. The rise of Large Language Models (LLMs) offers a new paradigm for ABM, enabling high-fidelity modeling of human behavior. However, urban dynamics often require extensive agent interactions to emerge, and scaling up LLM agent simulations is limited by high-latency remote LLM inference and high costs. To address this, we propose the OpenCity framework for large-scale LLM agent simulation. We design a novel "Group-and-Distill" strategy that significantly reduces costs without sacrificing behavioral diversity by aggregating agents with shared static characteristics while preserving their unique dynamic states. Moreover, a system-level request scheduler is designed to optimize I/O concurrency and communication bottlenecks. Experiments in six cities globally demonstrate that OpenCity achieves a 600x speedup and a 45% reduction in token usage compared to standard baseline models, while maintaining 96% behavioral consistency with independently invoked agent baselines. Based on OpenCity, we conduct first benchmark test on large-scale social segregation with LLM agents. The results show that OpenCity-accelerated generative agents are effective in reproducing socioeconomic segregation patterns. Furthermore, our scalability analysis demonstrates that the realism of emerging social phenomena increases as the agents scale up, quantitatively validating the necessity of large-scale simulations. Finally, we demonstrate the framework's capability as a policy sandbox through a counterfactual analysis of urban equity. Overall, OpenCity provides a robust and efficient framework that unlocks the potential of LLMs for large-scale, high-fidelity multi-agent systems analysis. Code repo is available at https://github.com/tsinghua-fib-lab/OpenCity.
Qingbin Zeng, Yuwei Yan, Zhiheng Zheng et al.· Proceedings of the 32nd ACM...· 0 citations
FlashRT is presented, an agent harness that guides coding agents to lift simple developer-written reference implementations into optimized multi-GPU deployments that flexibly weigh target metrics like latency and throughput, demonstrating that agent-driven optimization can be more scalable on platforms with less mature expert optimization.
Krish Agarwal, Zhuoming Chen, Yanyuan Qin et al.· 0 citations
A solver-grounded design principle is presented: a numerical result is reported only when it originates from a trusted tool and passes explicit verification, and a four-group evaluation framework spanning task utility, solver-grounded correctness, faithfulness and safe failure, and cost and latency is proposed.
Daniel Rojas, Abdulwahab Albassam, Aidan G. Leung et al.· 0 citations