# RealUserSim: Bridging the Reality Gap in Agent Benchmarking via Grounded User Simulation

arXiv: https://arxiv.org/abs/2605.20204 · HTML: https://mlsanity.com/paper/2605.20204/

## Key facts

- Cited by 12 papers in the map (data through 14 Oct 2026).
- Placed in Social simulation and multi-agent debate, one of 16 directions in LLM agents and tool use.
- Builds on 31 of its 34 references that are in the map.
- First posted on arXiv 7 Apr 2026.

## Placed in

- LLM agents and tool use / [Social simulation and multi-agent debate](https://mlsanity.com/agents/social-simulation-and-multi-agent-debate/)

## Builds on (most cited first, 30 of 31)

- [Generative Agents: Interactive Simulacra of Human Behavior](https://mlsanity.com/paper/2304.03442/) (2023)
- [τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domains](https://mlsanity.com/paper/2406.12045/) (2024)
- [Personalizing Dialogue Agents: I have a dog, do you have pets too?](https://mlsanity.com/paper/1801.07243/) (2018)
- [WildChat: 1M ChatGPT Interaction Logs in the Wild](https://mlsanity.com/paper/2405.01470/) (2024)
- [τ2-Bench: Evaluating Conversational Agents in a Dual-Control Environment](https://mlsanity.com/paper/2506.07982/) (2025)
- [Character-LLM: A Trainable Agent for Role-Playing](https://mlsanity.com/paper/2310.10158/) (2023)
- [Role-Play with Large Language Models](https://mlsanity.com/paper/2305.16367/) (2023)
- [From Persona to Personalization: A Survey on Role-Playing Language Agents](https://mlsanity.com/paper/2404.18231/) (2024)
- [SWEET-RL: Training Multi-Turn LLM Agents on Collaborative Reasoning Tasks](https://mlsanity.com/paper/2503.15478/) (2025)
- [PersonaGym: Evaluating Persona Agents and LLMs](https://mlsanity.com/paper/2407.18416/) (2024)
- [Mind the Sim2Real Gap in User Simulation for Agentic Tasks](https://mlsanity.com/paper/2603.11245/) (2026)
- [Neural User Simulation for Corpus-based Policy Optimisation for Spoken Dialogue Systems](https://mlsanity.com/paper/1805.06966/) (2018)
- [Know You First and Be You Better: Modeling Human-Like User Simulators via Implicit Profiles](https://mlsanity.com/paper/2502.18968/) (2025)
- [Collab-Overcooked: Benchmarking and Evaluating Large Language Models as Collaborative Agents](https://mlsanity.com/paper/2502.20073/) (2025)
- [Lost in Simulation: LLM-Simulated Users are Unreliable Proxies for Human Users in Agentic Evaluations](https://mlsanity.com/paper/2601.17087/) (2026)
- [Consistently Simulating Human Personas with Multi-Turn Reinforcement Learning](https://mlsanity.com/paper/2511.00222/) (2025)
- [AgentLAB: Benchmarking LLM Agents against Long-Horizon Attacks](https://mlsanity.com/paper/2602.16901/) (2026)
- [SimUSER: Simulating User Behavior with Large Language Models for Recommender System Evaluation](https://mlsanity.com/paper/2504.12722/) (2025)
- [BASES: Large-scale Web Search User Simulation with Large Language Model based Agents](https://mlsanity.com/paper/2402.17505/) (2024)
- [AgentRecBench: Benchmarking LLM Agent-based Personalized Recommender Systems](https://mlsanity.com/paper/2505.19623/) (2025)
- [RealMem: Benchmarking LLMs in Real-World Memory-Driven Interaction](https://mlsanity.com/paper/2601.06966/) (2026)
- [AMemGym: Interactive Memory Benchmarking for Assistants in Long-Horizon Conversations](https://mlsanity.com/paper/2603.01966/) (2026)
- [DeepPersona: A Generative Engine for Scaling Deep Synthetic Personas](https://mlsanity.com/paper/2511.07338/) (2025)
- [InteractComp: Evaluating Search Agents With Ambiguous Queries](https://mlsanity.com/paper/2510.24668/) (2025)
- [IDA-Bench: Evaluating LLMs on Interactive Guided Data Analysis](https://mlsanity.com/paper/2505.18223/) (2025)
- [ProductAgent: Benchmarking Conversational Product Search Agent with Asking Clarification Questions](https://mlsanity.com/paper/2407.00942/) (2024)
- [Effective Red-Teaming of Policy-Adherent Agents](https://mlsanity.com/paper/2506.09600/) (2025)
- [HER: Human-like Reasoning and Reinforcement Learning for LLM Role-playing](https://mlsanity.com/paper/2601.21459/) (2026)
- [SAGE: A Top-Down Bottom-Up Knowledge-Grounded User Simulator for Multi-turn AGent Evaluation](https://mlsanity.com/paper/2510.11997/) (2025)
- [Implicit Behavioral Alignment of Language Agents in High-Stakes Crowd Simulations](https://mlsanity.com/paper/2509.16457/) (2025)

## Built on by (most cited first, 12 of 12)

- [Beyond Cooperative Simulators: Generating Realistic User Personas for Robust Evaluation of LLM Agents](https://mlsanity.com/paper/2605.12894/) (2026)
- [VISTA: A Versatile Interactive User Simulation Toolkit for Agent Evaluation](https://mlsanity.com/paper/2606.11079/) (2026)
- [Simulated Customers Never Walk Away: Decision Fidelity of LLM User Simulators Measured Against Real Purchase Outcomes](https://mlsanity.com/paper/2606.20708/) (2026)
- [BehaviorBench: Modeling Real-World User Decisions from Behavioral Traces](https://mlsanity.com/paper/2606.02798/) (2026)
- [Investigating Assistant Bias in LLM User Simulators Using a Role Vector](https://mlsanity.com/paper/2609.00608/) (2026)
- [Screen Before You Serve: Simulation for Production Customer Experience AI Agents at 140M Scale](https://mlsanity.com/paper/2609.30137/) (2026)
- [On Evaluating and Improving Conversational Agents in Production](https://mlsanity.com/paper/2609.32092/) (2026)
- [CUEing User Simulators: Calibrated User Embeddings for Multi-Turn Benchmarking](https://mlsanity.com/paper/2610.02460/) (2026)
- [MIMESIS: Learning User Simulators as Training Environments for Interactive Agents](https://mlsanity.com/paper/2610.09484/) (2026)
- [Training User Simulators with Latent Intrinsic Feedback](https://mlsanity.com/paper/2610.13617/) (2026)
- [Uncovering Robustness Weaknesses in LLM Agents via Persona Evolution with Task-Integrity Constraints](https://mlsanity.com/paper/2610.16155/) (2026)
- [Cowork-Gym: Synthesizing Interactive Work Environments at Scale](https://mlsanity.com/paper/2610.17670/) (2026)

## Abstract (arXiv)

LLM-based user simulation is the primary mechanism for end-to-end agent evaluation, yet simulated users are poor proxies for real humans: unconstrained LLM defaults produce a Formalism Ceiling (style match rates of 6-8% against real users), while hand-crafted behavioral directives trigger Directive Amplification, where models hyper-interpret instructions into unnatural behavioral extremes that vary dramatically across simulator models. We present RealUserSim, the first user simulation framework grounded in real behavioral data. From 14,000+ authentic human-LLM conversations (WildChat), we extract 7,275 executable behavioral profiles and use them to ground LLM simulators. A fidelity benchmark (PT3) on 600 conversations across 71+ domains with anti-leakage controls shows that grounded simulation raises match rate from 24.2% to 45.3% across five behavioral dimensions. Agent evaluation on TauBench with 6 simulator models and extensive analysis shows that grounded simulation acts as a realistic stress test, surfacing three failure mechanisms invisible to cooperative simulators (mean -3.2% to -3.5% task success degradation), while Directive Amplification in existing benchmarks produces unrealistic behavior that compromises the validity of agent evaluation.