AGENTSERVESIM: Hardware-aware simulator for multi-turn LLM agent serving policies
Researchers introduce AGENTSERVESIM, a simulation framework designed to evaluate serving policies for multi-turn LLM agents without requiring dedicated accelerator hardware. The simulator models program-level execution including turn dependencies, tool-induced gaps, and KV-cache residency across HBM, host DRAM, and CXL memory hierarchies. It reproduces real-system behavior within 6% error on key performance metrics while running on commodity CPUs, enabling cost-effective exploration of scheduling, routing, and cache management policies for agentic workloads.
Related guides (3)
Related events (8)
Agentopia: Long-term multi-agent life simulation framework for training LLMs on social behavior
Researchers introduce Agentopia, a framework for simulating 10 years of social life across 100 LLM-powered agents, enabling study of emergent social behaviors and long-term personal growth dynamics. The system defines a 'life reward' metric mirroring human well-being and uses it to train LLMs via rejection sampling. Training on simulated social experience yields a +15.6% improvement on downstream role-playing benchmarks, suggesting that synthetic social simulation can generalize to real capability gains.
CollabSim: CSCW-grounded framework for evaluating collaborative competence in LLM multi-agent systems
Researchers introduce CollabSim, a configurable simulation framework for systematically evaluating collaborative competence in LLM-based multi-agent systems (MAS). The framework draws on Computer-Supported Cooperative Work (CSCW) theory to define collaborative capabilities beyond task outcomes, including common ground establishment, shared task understanding, and misalignment repair. Experiments across four LLMs demonstrate the framework can distinguish model performance patterns and reveal task-dependent effects of agent design choices. The work addresses a gap in MAS evaluation, which has historically focused on individual task-solving rather than coordination quality.
SIMMER benchmark exposes high rates of latent planning failures in frontier LLMs
Researchers introduce SIMMER, a benchmark for evaluating latent failures in LLM-generated executable plans within a kitchen-domain world model comprising 77 actions, 262 objects, and ~46,800 possible interactions. Unlike existing benchmarks that only catch immediate execution failures, SIMMER detects silent hazards and irreversible consequences using a state machine executor. Experiments across six LLMs find that even frontier models produce error-free plans at most 17% of the time, with up to 56% of plans containing latent failures—most leading to irreversible outcomes. The paper also shows that counterfactual foresight simulation can reduce latent failures by up to 72%, pointing toward a mitigation direction.
AgentSpec: A modular framework for controlled composition and analysis of embodied LLM agent scaffolds
AgentSpec is a new modular specification framework that represents embodied LLM agents as typed compositions of reusable policy components with standardized interfaces across perception, memory, reasoning, reflection, action, and learning modules. The framework enables controlled swapping and recombination of components, instantiated across four benchmarks (DeliveryBench, ALFRED, MiniGrid, RoboTHOR). Key findings include that agent performance is governed by scaffold compatibility and interaction effects rather than isolated module strength, and that RL-trained policies compose best when optimized with deployment-time scaffold structure. Code, baselines, and an interactive playground are publicly released.
MobileGym: Verifiable Parallel Simulation Platform for Mobile GUI Agent Training
MobileGym is a browser-hosted simulation environment for mobile GUI agent research that enables deterministic outcome verification via structured JSON state and scalable online RL through hundreds of parallel instances (~400 MB/instance, ~3s cold start). The accompanying MobileGym-Bench provides 416 parameterized task templates across 28 apps with deterministic judges. A sim-to-real case study using GRPO on Qwen3-VL-4B-Instruct achieves +12.8 percentage points on the 256-task test set, with real-device execution retaining 95.1% of simulation-side training gains.
AgentMob: Training-free LLM agent framework for evidence-grounded mobility prediction
AgentMob is a training-free LLM-driven agent framework that formulates next-location prediction as adaptive evidence-controlled decision making, using a fast path for routine cases and iterative tool use for ambiguous ones. Evaluated on three mobility datasets, it achieves the strongest overall performance among training-free LLM-based methods, with GPT-5.4 reaching 71.42% Acc@1 on the BW dataset. The framework demonstrates that LLM controllers add most value in resolving ambiguous predictions through adaptive evidence gathering rather than routine cases.
Multi-agent LLM framework for Chinese civil court simulation with five-stage trial procedure
Researchers present a multi-agent LLM framework for simulating Chinese civil court proceedings, organized around a five-stage civil trial procedure with memory modules and statute retrieval. The system targets civil litigation specifically, which is more common and harder to simulate than criminal cases due to flexible claims and remedies. Experiments show reliable judgment outputs with particular strengths in liability allocation, and find that memory quality substantially affects downstream simulation quality. Code and dataset are publicly released.
LongMINT: Benchmark for Evaluating Memory Under Multi-Target Interference in Long-Horizon Agent Systems
LongMINT is a new benchmark designed to evaluate memory-augmented agents in realistic long-horizon settings where information is repeatedly updated and interferes across memories. It contains 15.6k QA pairs over contexts averaging 138.8k tokens (up to 1.8M tokens), spanning domains including state tracking, multi-turn dialogue, Wikipedia revisions, and GitHub commits. Evaluation of 7 representative systems—including vanilla long-context LLMs, RAG, and memory-augmented agent frameworks—reveals consistently low average accuracy of 27.9%, with performance particularly degraded on multi-target aggregation tasks and when earlier facts are revised by subsequent context. The analysis identifies retrieval and memory construction as the primary bottlenecks.