Asaf Yehudai, Lilach Eden, Alan Li, Guy Uziel, Yilun Zhao, Roy Bar-Haim, Arman Cohan, Michal Shmueli-Scheuer
| Challenge: | This paper provides the first comprehensive survey of evaluation methods for LLM-based agents . LLMs are static, having fixed knowledge, and confined to text-to-text interaction. |
| Approach: | They analyze the evaluation of LLM-based agents across five perspectives . they identify current trends and key gaps in evaluation methods . |
| Outcome: | The proposed evaluation frameworks and tools are based on five perspectives . the results highlight current trends and identify gaps in future research . |
Similar Papers
An Evaluation Mechanism of LLM-based Agents on Manipulating APIs (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have remarkable capabilities across a variety of tasks, such as language, mathematics, coding, and etc. |
| Approach: | They propose to decompose tool use capability into seven aspects and form a thorough evaluation schema for generic agents. |
| Outcome: | The proposed agent acts like a super-APP and can manipulate API-based tools. |
PersonaGym: Evaluating Persona Agents and LLMs (2025.findings-emnlp)
Copied to clipboard
Vinay Samuel, Henry Peng Zou, Yue Zhou, Shreyas Chaudhari, Ashwin Kalyan, Tanmay Rajpurohit, Ameet Deshpande, Karthik R Narasimhan, Vishvak Murahari
| Challenge: | Persona agents are LLM agents conditioned to act according to an assigned persona . evaluating how faithfully these agents adhere to their personas remains a challenge . |
| Approach: | a new study evaluates persona agents' ability to act according to an assigned persona . a persona agent's person score is a human-aligned automatic metric that can be used to evaluate a model . |
| Outcome: | a new evaluation framework and a human-aligned automatic metric show that persona agents can perform better. |
Process Evaluation for Agentic Systems (2026.findings-eacl)
Copied to clipboard
| Challenge: | Recent adoption of LLM-based assistants has led to premature assumptions about their reliability and general capability. |
| Approach: | They propose to assess the feasibility of automatic process evaluation for critical applications such as medicine, finance, law and infrastructure. |
| Outcome: | The proposed evaluations are based on a small-scale study to assess the feasibility of automated process evaluation, present a compliance score, analyse use cases of bad and good behaviours, and offer recommendations for more holistic evaluation. |
Mobile-Bench: An Evaluation Benchmark for LLM-based Mobile Agents (2024.acl-long)
Copied to clipboard
Shihan Deng, Weikai Xu, Hongda Sun, Wei Liu, Tao Tan, Liujianfeng Liujianfeng, Ang Li, Jian Luan, Bin Wang, Rui Yan, Shuo Shang
| Challenge: | Existing benchmarks for LLM-based mobile agents are insufficient to evaluate their capabilities. |
| Approach: | They propose a benchmark to evaluate LLM-based mobile agents' planning capabilities . they expand UI operations by incorporating 103 APIs to accelerate task completion . |
| Outcome: | The proposed benchmarks are based on 103 collected APIs and real user queries . the data is categorized into three distinct groups: SAST, SAMT, and MAMT . |
AgentGym2: Benchmarking Large Language Model Agents in De-Idealized Real-World Environments (2026.acl-long)
Copied to clipboard
Zhiheng Xi, Dingwen Yang, Jiaqi Liu, Jixuan Huang, Honglin Guo, Baodai Huang, Tinggang Chen, Qi Zhang, Zhonghang Lu, Chenyu Liu, Jiajun Sun, Jiazheng Zhang, Dingwei Zhu, Xin Guo, Junzhe Wang, Zhihao Zhang, Yuming Yang, Junjie Ye, Minghe Gao, Dongrui Liu, Jiaming Ji, Guohao Li, Tao Gui, Qi Zhang, Xuanjing Huang
| Challenge: | Existing benchmarks evaluate agents in simplified, idealized settings, relying on pre-packaged tool interfaces, overlooking critical steps, and assume inputs are clean and fully specified. |
| Approach: | They propose a framework that evaluates language agents in simplified, idealized settings . they show that even SOTA systems like Gemini and GPT-5 struggle on AgentGym2 . |
| Outcome: | Experiments on 15 proprietary and open-source models show that even SOTA systems like Gemini and GPT-5 struggle on AgentGym2 . |
Agentic CLEAR: Automating Multi-Level Evaluation of LLM Agents (2026.acl-demo)
Copied to clipboard
| Challenge: | Agentic systems are becoming more capable of defining strategies, taking actions, and solving complex, multi-step tasks. |
| Approach: | They propose an automatic, dynamic, and easy-to-use evaluation framework that provides textual insights into agent behavior on three levels of granularity: system, trace, and node. |
| Outcome: | The proposed framework produces high-quality, data-driven, insightful feedback on system, trace, and node. |
AgencyBench: Benchmarking the Frontiers of Autonomous Agents in 1M-Token Real-World Contexts (2026.acl-long)
Copied to clipboard
Keyu Li, Junhao Shi, Yang Xiao, Mohan Jiang, Jie Sun, Yunze Wu, Dayuan Fu, Shijie Xia, Xiaojie Cai, Tianze Xu, Weiye Si, Wenjie Li, Dequan Wang, Pengfei Liu
| Challenge: | Existing benchmarks focus on single agentic capability, failing to capture long-horizon real-world scenarios. |
| Approach: | They propose a benchmark that evaluates 6 agentic capabilities across 32 real-world scenarios. |
| Outcome: | Experiments show that closed-source models outperform open-source model (48.4% vs 32.1%) integrating models with advanced scaffolds to form autonomous agents is a paradigm shift. |
AgentGym: Evaluating and Training Large Language Model-based Agents across Diverse Environments (2025.acl-long)
Copied to clipboard
Zhiheng Xi, Yiwen Ding, Wenxiang Chen, Boyang Hong, Honglin Guo, Junzhe Wang, Xin Guo, Dingwen Yang, Chenyang Liao, Wei He, Songyang Gao, Lu Chen, Rui Zheng, Yicheng Zou, Tao Gui, Qi Zhang, Xipeng Qiu, Xuanjing Huang, Zuxuan Wu, Yu-Gang Jiang
| Challenge: | Large language models (LLMs) are promising foundations to build generally-capable agents . however, the community lacks a unified interactive framework that covers diverse environments for comprehensive evaluation of agents. |
| Approach: | They propose a framework that features 7 real-world scenarios, 14 environments, and 89 tasks for unified, real-time, and concurrent agent interaction. |
| Outcome: | The proposed framework features 7 real-world scenarios, 14 environments, and 89 tasks for unified, real-time, and concurrent agent interaction. |
A Review of Prominent Paradigms for LLM-Based Agents: Tool Use, Planning (Including RAG), and Feedback Learning (2025.coling-main)
Copied to clipboard
| Challenge: | Large Language Models have been used for planning, tool use, and feedback learning . inconsistent taxonomy and complexity of workflows create challenges . |
| Approach: | They propose a unified taxonomy to review and discuss the three paradigms . they define environments/tasks, common LLM-profiled roles and universally applicable workflows based on prior work . |
| Outcome: | The proposed taxonomy compares LMPR implementations and workflow usage across paradigms . large language models have human-like reasoning capabilities, the authors say . |
Can LLMs Help You at Work? A Sandbox for Evaluating LLM Agents in Enterprise Environments (2025.emnlp-main)
Copied to clipboard
| Challenge: | Enterprise systems are crucial for enhancing productivity and strategic growth, but data is fragmented across multiple sources and access controls are complex. |
| Approach: | They propose a benchmark that simulates enterprise settings with 500 diverse tasks . they show that even the most capable models achieve only 41.8% task completion . |
| Outcome: | The proposed benchmark shows that even the most capable models achieve only 41.8% task completion. |