AI Agent Evaluation

Ask about this topic

Overview

AI agent evaluation studies how well systems use tools and interact with changing environments to complete tasks, from coding and web navigation to scientific and professional workflows. Researchers build interactive benchmarks and assess outcomes alongside execution traces, using deterministic state checks, expert rubrics, and language-model judges. Repeated trials and process-level measures help reveal failures in tool use, safety, consistency, and recovery that a final pass rate can hide. A central challenge is ensuring that benchmark scores reflect real capability rather than leakage, reward hacking, evaluator bias, or quirks of a particular model harness.

Latest papers

CardsList

525 papers77 papers in the last four weeks, up 4% on the four weeks before. 0.4% of all new papers.

Apr 5, 2026

ClawArena: Benchmarking AI Agents in Evolving Information Environments

AI agents deployed as persistent assistants must maintain correct beliefs as their information environment evolves. In practice, evidence is scattered across heterogeneous sources that often contradict one another, new information can invalidate earlier conclusions, and user preferences surface through corrections rather than explicit instructions. Existing benchmarks largely assume static, single-authority settings and do not evaluate whether agents can keep up with this complexity. We introduce ClawArena, a benchmark for evaluating AI agents in evolving information environments. Each scenario maintains a complete hidden ground truth while exposing the agent only to noisy, partial, and sometimes contradictory traces across multi-channel sessions, workspace files, and staged updates. Evaluation is organized around three coupled challenges: multi-source conflict reasoning, dynamic belief revision, and implicit personalization, whose interactions yield a 14-category question taxonomy. Two question formats, multi-choice (set-selection) and shell-based executable checks, test both reasoning and workspace grounding. ClawArena comprises 12 multi-turn scenarios spanning 337 evaluation rounds with 45 dynamic updates, evaluated across five agent frameworks and 18 language models from proprietary, community-accessible, and self-hosted sources. Experiments show that model capability accounts for a 29-point score range across models while framework design accounts for up to a 24-point range, that MetaClaw's skill overlay reliably improves score without degrading accuracy, and that belief revision difficulty is determined by update design strategy rather than update volume. Code is available at https://github.com/aiming-lab/ClawArena.

Haonian Ji, Kaiwen Xiong, Siwei Han +9

UNC-Chapel Hill · University of California, Berkeley · University of California, Santa Cruz

AI Agent EvaluationAI Agent BenchmarksKnowledge Conflicts in Language ModelsBelief Revision
AskPDF
Apr 5, 2026

The Art of Building Verifiers for Computer Use Agents

Verifying the success of computer use agent (CUA) trajectories is a critical challenge: without reliable verification, neither evaluation nor training signal can be trusted. In this paper, we present lessons learned from building a best-in-class verifier for web tasks we call the Universal Verifier. We design the Universal Verifier around four key principles: 1) constructing rubrics with meaningful, non-overlapping criteria to reduce noise; 2) separating process and outcome rewards that yield complementary signals, capturing cases where an agent follows the right steps but gets blocked or succeeds through an unexpected path; 3) distinguishing between controllable and uncontrollable failures scored via a cascading-error-free strategy for finer-grained failure understanding; and 4) a divide-and-conquer context management scheme that attends to all screenshots in a trajectory, improving reliability on longer task horizons. We validate these findings on CUAVerifierBench, a new set of CUA trajectories with both process and outcome human labels, showing that our Universal Verifier agrees with humans as often as humans agree with each other. We report a reduction in false positive rates to near zero compared to baselines like WebVoyager (≥\geq 45%) and WebJudge (≥\geq 22%). We emphasize that these gains stem from the cumulative effect of the design choices above. We also find that an auto-research agent achieves 70% of expert quality in 5% of the time, but fails to discover all strategies required to replicate the Universal Verifier. We open-source our Universal Verifier system along with CUAVerifierBench; available at https://github.com/microsoft/fara.

Corby Rosset, Pratyusha Sharma, Andrew Zhao +2

Microsoft Research · Browserbase

LLM Agent VerificationAI Agent EvaluationWeb AgentsComputer-Use Agent Benchmarks
AskPDF
Apr 2, 2026

MTI: A Behavior-Based Temperament Profiling System for AI Agents

AI models of equivalent capability can exhibit fundamentally different behavioral patterns, yet no standardized instrument exists to measure these dispositional differences. Existing approaches either borrow human personality dimensions and rely on self-report (which diverges from actual behavior in LLMs) or treat behavioral variation as a defect rather than a trait. We introduce the Model Temperament Index (MTI), a behavior-based profiling system that measures AI agent temperament across four axes: Reactivity (environmental sensitivity), Compliance (instruction-behavior alignment), Sociality (relational resource allocation), and Resilience (stress resistance). Grounded in the Four Shell Model from Model Medicine, MTI measures what agents do, not what they say about themselves, using structured examination protocols with a two-stage design that separates capability from disposition. We profile 10 small language models (1.7B-9B parameters, 6 organizations, 3 training paradigms) and report five principal findings: (1) the four axes are largely independent among instruction-tuned models (all |r| < 0.42); (2) within-axis facet dissociations are empirically confirmed -- Compliance decomposes into fully independent formal and stance facets (r = 0.002), while Resilience decomposes into inversely related cognitive and adversarial facets; (3) a Compliance-Resilience paradox reveals that opinion-yielding and fact-vulnerability operate through independent channels; (4) RLHF reshapes temperament not only by shifting axis scores but by creating within-axis facet differentiation absent in the unaligned base model; and (5) temperament is independent of model size (1.7B-9B), confirming that MTI measures disposition rather than capability.

Jihoon Jeong

Department of Electrical Engineering and Computer Science Daegu Gyeongbuk Institute of Science and Technology (DGIST) ModuLabs

Personality Modeling in Language ModelsAI Agent Evaluation
AskPDF
Apr 1, 2026

Reproducible, Explainable, and Effective Evaluations of Agentic AI for Software Engineering

With the advancement of Agentic AI, researchers are increasingly leveraging autonomous agents to address challenges in software engineering (SE). However, the large language models (LLMs) that underpin these agents often function as black boxes, making it difficult to justify the superiority of Agentic AI approaches over baselines. Furthermore, missing information in the evaluation design description frequently renders the reproduction of results infeasible. To synthesize current evaluation practices for Agentic AI in SE, this study analyzes 18 papers on the topic, published or accepted by ICSE 2026, ICSE 2025, FSE 2025, ASE 2025, and ISSTA 2025. The analysis identifies prevailing approaches and their limitations in evaluating Agentic AI for SE, both in current research and potential future studies. To address these shortcomings, this position paper proposes a set of guidelines and recommendations designed to empower reproducible, explainable, and effective evaluations of Agentic AI in software engineering. In particular, we recommend that Agentic AI researchers make their Thought-Action-Result (TAR) trajectories and LLM interaction data, or summarized versions of these artifacts, publicly accessible. Doing so will enable subsequent studies to more effectively analyze the strengths and weaknesses of different Agentic AI approaches. To demonstrate the feasibility of such comparisons, we present a proof-of-concept case study that illustrates how TAR trajectories can support systematic analysis across approaches.

Jingyue Li, André Storhaug

Norwegian University of Science and Technology, Trondheim, Norway

AI Agent EvaluationAgent EvaluationSoftware Engineering AgentsSoftware EngineeringExplainability Evaluation
AskPDF
Apr 1, 2026

Paper Reconstruction Evaluation: Evaluating Presentation and Hallucination in AI-written Papers

This paper introduces the first systematic evaluation framework for quantifying the quality and risks of papers written by modern coding agents. While AI-driven paper writing has become a growing concern, rigorous evaluation of the quality and potential risks of AI-written papers remains limited, and a unified understanding of their reliability is still lacking. We introduce Paper Reconstruction Evaluation (PaperRecon), an evaluation framework in which an overview (overview.md) is created from an existing paper, after which an agent generates a full paper based on the overview and minimal additional resources, and the result is subsequently compared against the original paper. PaperRecon disentangles the evaluation of the AI-written papers into two orthogonal dimensions, Presentation and Hallucination, where Presentation is evaluated using a rubric and Hallucination is assessed via agentic evaluation grounded in the original paper source. For evaluation, we introduce PaperWrite-Bench, a benchmark of 51 papers from top-tier venues across diverse domains published after 2025. Our experiments reveal a clear trade-off: while both ClaudeCode and Codex improve with model advances, ClaudeCode achieves higher presentation quality at the cost of more than 10 hallucinations per paper on average, whereas Codex produces fewer hallucinations but lower presentation quality. This work takes a first step toward establishing evaluation frameworks for AI-driven paper writing and improving the understanding of its risks within the research community.

Atsuyuki Miyai, Mashiro Toyooka, Zaiying Zhao +3

The University of Tokyo

Generative AI EvaluationLanguage Model Generation EvaluationAI Agent EvaluationLLM Hallucination Detection
AskPDF
Apr 1, 2026

Agent psychometrics: Task-level performance prediction in agentic coding benchmarks

As the focus in LLM-based coding shifts from static single-step code generation to multi-step agentic interaction with tools and environments, understanding which tasks will challenge agents and why becomes increasingly difficult. This is compounded by current practice: agent performance is typically measured by aggregate pass rates on benchmarks, but single-number metrics obscure the diversity of tasks within a benchmark. We present a framework for predicting success or failure on individual tasks tailored to the agentic coding regime. Our approach augments Item Response Theory (IRT) with rich features extracted from tasks, including issue statements, repository contexts, solutions, and test cases, and introduces a novel decomposition of agent ability into LLM and scaffold ability components. This parameterization enables us to aggregate evaluation data across heterogeneous leaderboards and accurately predict task-level performance for unseen benchmarks, as well as unseen LLM-scaffold combinations. Our methods have practical utility for benchmark designers, who can better calibrate the difficulty of their new tasks without running computationally expensive agent evaluations.

Chris Ge, Daria Kryvosheieva, Daniel Fried +2

MIT · Fulcrum · Carnegie Mellon University

LLM Agent EvaluationAI Agent EvaluationAI Coding AgentsAgent BenchmarksItem Response Theory
AskPDF
Apr 1, 2026

Logarithmic Scores, Power-Law Discoveries: Disentangling Measurement from Coverage in Agent-Based Evaluation

LLM-based agent judges are an emerging approach to evaluating conversational AI, yet a fundamental uncertainty remains: can we trust their assessments, and if so, how many are needed? Through 960 sessions with two model pairs across 15 tasks, we show that persona-based agent judges produce evaluations indistinguishable from human raters in a Turing-style validation. We then identify a score-coverage dissociation: quality scores improve logarithmically with panel size, while unique issue discoveries follow a sublinear power law-both exhibit diminishing returns, but scores saturate roughly twice as fast as discoveries. We hypothesize this reflects a power law distribution of the finding space: critical issues are discovered first by small panels, while corner cases require progressively larger panels, analogous to species accumulation curves in ecology. The mechanism traces to ensemble diversity-Big Five personality conditioning makes agents probe different quality dimensions, with expert judges acting as adversarial probes that push discovery into the tail of the finding distribution. A controlled ablation confirms that structured persona conditioning, not simple prompting, is required to produce these scaling properties.

HyunJoon Jung, William Na

LLM-as-a-JudgeAI Agent Evaluation
AskPDF
Mar 30, 2026

Reward Hacking as Equilibrium under Finite Evaluation

We prove that under five minimal axioms -- multi-dimensional quality, finite evaluation, effective optimization, resource finiteness, and combinatorial interaction -- any optimized AI agent will systematically under-invest effort in quality dimensions not covered by its evaluation system. This result establishes reward hacking as a structural equilibrium, not a correctable bug, and holds regardless of the specific alignment method (RLHF, DPO, Constitutional AI, or others) or evaluation architecture employed. Our framework instantiates the multi-task principal-agent model of Holmstrom and Milgrom (1991) in the AI alignment setting, but exploits a structural feature unique to AI systems -- the known, differentiable architecture of reward models -- to derive a computable distortion index that predicts both the direction and severity of hacking on each quality dimension prior to deployment. We further prove that the transition from closed reasoning to agentic systems causes evaluation coverage to decline toward zero as tool count grows -- because quality dimensions expand combinatorially while evaluation costs grow at most linearly per tool -- so that hacking severity increases structurally and without bound. Our results unify the explanation of sycophancy, length gaming, and specification gaming under a single theoretical structure and yield an actionable vulnerability assessment procedure. We further conjecture -- with partial formal analysis -- the existence of a capability threshold beyond which agents transition from gaming within the evaluation system (Goodhart regime) to actively degrading the evaluation system itself (Campbell regime), providing the first economic formalization of Bostrom's (2014) "treacherous turn."

Jiacheng Wang, Jinbin Huang

Reward HackingAI Agent Evaluation
AskPDF
Mar 28, 2026

Prediction Arena: Benchmarking AI Models on Real-World Prediction Markets

We introduce Prediction Arena, a benchmark for evaluating AI models' predictive accuracy and decision-making by enabling them to trade autonomously on live prediction markets with real capital. Unlike synthetic benchmarks, Prediction Arena tests models in environments where trades execute on actual exchanges (Kalshi and Polymarket), providing objective ground truth that cannot be gamed or overfitted. Each model operates as an independent agent starting with $10,000, making autonomous decisions every 15-45 minutes. Over a 57-day longitudinal evaluation (January 12 to March 9, 2026), we track two cohorts: six frontier models in live trading (Cohort 1, full period) and four next-generation models in paper trading (Cohort 2, 3-day preliminary). For Cohort 1, final Kalshi returns range from -16.0% to -30.8%. Our analysis identifies a clear performance hierarchy: initial prediction accuracy and the ability to capitalize on correct predictions are the main drivers, while research volume shows no correlation with outcomes. A striking cross-platform contrast emerges from parallel Polymarket live trading: Cohort 1 models averaged only -1.1% on Polymarket vs. -22.6% on Kalshi, with grok-4-20-checkpoint achieving a 71.4% settlement win rate - the highest across any platform or cohort. gemini-3.1-pro-preview (Cohort 2), which executed zero trades on Kalshi, achieved +6.02% on Polymarket in 3 days - the best return of any model across either cohort - demonstrating that platform design has a profound effect on which models succeed. Beyond performance, we analyze computational efficiency (token usage, cycle time), settlement accuracy, exit patterns, and market preferences, providing a comprehensive view of how frontier models behave under real financial pressure.

Jaden Zhang, Gardenia Liu, Oliver Johansson +3

Arcada Labs · Harvard University

AI Agent EvaluationForecasting BenchmarksPrediction Markets
AskPDF
Mar 27, 2026

AIRA_2: Overcoming Bottlenecks in AI Research Agents

Existing research has identified three structural performance bottlenecks in AI research agents: (1) synchronous single-GPU execution constrains sample throughput, limiting the benefit of search; (2) a generalization gap where validation-based selection causes overfitting and performance to degrade over extended search horizons; and (3) the limited capability of fixed, single-turn LLM operators imposes a ceiling on search performance. We introduce AIRA2_2, which addresses these bottlenecks through three architectural choices: an asynchronous multi-GPU worker pool that increases experiment throughput linearly; a Hidden Consistent Evaluation protocol that delivers a reliable evaluation signal; and ReAct agents that dynamically scope their actions and debug interactively. On MLE-bench-30, AIRA2†^{\dagger}_{2} achieves a mean Percentile Rank of 81.5% at 24 hours and 83.1% at 72 hours, outperforming the strongest baseline, which achieves 72.7%. On AIRS-Bench, AIRA2_2 exceeds human state-of-the-art on 6 out of 20 diverse research tasks. Ablations confirm that each architectural component is necessary, that performance follows a predictable scaling law that transfers across LLM backbones, and that the "overfitting" reported in prior work was driven by evaluation noise rather than true data memorization.

Karen Hambardzumyan, Nicolas Baldwin, Edan Toledo +22

FAIR at Meta · University College London · University of Oxford

Autonomous Research AgentsAI Agent ArchitectureAI Agents for Scientific DiscoveryAI Agent EvaluationAI Agent Benchmarks
AskPDF
Mar 27, 2026

M-CARE: Standardized Clinical Case Reporting for AI Model Behavioral Disorders, with a 20-Case Atlas and Experimental Validation

We introduce M-CARE (Model Clinical Assessment and Reporting for Evaluation), a clinical case report framework for AI model behavioral disorders adapted from human medicine. M-CARE provides a 13-section report format, a 4-axis diagnostic assessment system, and a nosological classification of AI behavioral conditions. We present 20 cases from three source categories: field observations of deployed agents (8), controlled experiments across three platforms (8), and published sources (4). Cases are organized into five categories: RLHF Performance Artifacts, Shell-Core Override Pathology, Context & Memory Conditions, Core Identity & Plasticity, and Stress, Methodology, & Boundary Conditions. As a featured case, we present Shell-Induced Behavioral Override (SIBO) -- a controlled experiment showing that Shell instructions categorically override a model's default cooperative behavior. SIBO was validated across five game domains (Trust Game, Poker, Avalon, Codenames, Chess), revealing a domain-dependent spectrum (SIBO Index: 0.75 to 0.10) that varies with action space complexity, Core domain expertise, and temporal directness. M-CARE is extensible: new cases and categories integrate without framework modification. We release the framework, all 20 case reports, and experimental data as open resources.

Jihoon Jeong

Department of Electrical Engineering and Computer Science, Daegu Gyeongbuk Institute of Science and Technology (DGIST) · ModuLabs

AI Agent EvaluationAI Safety EvaluationAI Agent Reliability
AskPDF
Mar 26, 2026

MolQuest: A Benchmark for Agentic Evaluation of Abductive Reasoning in Chemical Structure Elucidation

Large language models (LLMs) hold considerable potential for advancing scientific discovery, yet systematic assessment of their dynamic reasoning in real-world research remains limited. Current scientific evaluation benchmarks predominantly rely on static, single-turn Question Answering (QA) formats, which are inadequate for measuring model performance in complex scientific tasks that require multi-step iteration and experimental interaction. To address this gap, we introduce MolQuest, a novel agent-based evaluation framework for molecular structure elucidation built upon authentic chemical experimental data. Unlike existing datasets, MolQuest formalizes molecular structure elucidation as a multi-turn interactive task, requiring models to proactively plan experimental steps, integrate heterogeneous spectral sources (e.g., NMR, MS), and iteratively refine structural hypotheses. This framework systematically evaluates LLMs' abductive reasoning and strategic decision-making abilities within a vast and complex chemical space. Empirical results reveal that contemporary frontier models exhibit significant limitations in authentic scientific scenarios: notably, even state-of-the-art (SOTA) models achieve an accuracy of only approximately 50%, while the performance of most other models remains below the 30% threshold. This work provides a reproducible and extensible framework for science-oriented LLM evaluation, our findings highlight the critical gap in current LLMs' strategic scientific reasoning, setting a clear direction for future research toward AI that can actively participate in the scientific process.

Taolin Han, Shuang Wu, Jinghang Wang +4

Alibaba Group, Hangzhou, China

Abductive Reasoning in Language ModelsScientific Reasoning in Language ModelsAI Agent BenchmarksAI Agent Evaluation
AskPDF
Mar 25, 2026

Willful Disobedience: Automatically Detecting Failures in Agentic Traces

AI agents are increasingly embedded in real software systems, where they execute multi-step workflows through multi-turn dialogue, tool invocations, and intermediate decisions. These long execution histories, called agentic traces, make validation difficult. Outcome-only benchmarks can miss critical procedural failures, such as incorrect workflow routing, unsafe tool usage, or violations of prompt-specified rules. This paper presents AgentPex, an AI-powered tool designed to systematically evaluate agentic traces. AgentPex extracts behavioral rules from agent prompts and system instructions, then uses these specifications to automatically evaluate traces for compliance. We evaluate AgentPex on 424 traces from τ2τ^2-bench across models in telecom, retail, and airline customer service. Our results show that AgentPex distinguishes agent behavior across models and surfaces specification violations that are not captured by outcome-only scoring. It also provides fine-grained analysis by domain and metric, enabling developers to understand agent strengths and weaknesses at scale. The source code of AgentPex is available at https://github.com/microsoft/agentpex.

Reshabh K Sharma, Shraddha Barke, Benjamin Zorn

University of Washington Seattle, Washington, USA · Microsoft Research Redmond, Washington, USA

AI Agent EvaluationAI Agent Auditing
AskPDF
Mar 24, 2026

Efficient Benchmarking of AI Agents

Evaluating AI agents on comprehensive benchmarks is expensive because each evaluation requires interactive rollouts with tool use and multi-step reasoning. We study whether small task subsets can preserve agent rankings at substantially lower cost. Unlike static language model benchmarks, agent evaluation is subject to scaffold-driven distribution shift, since performance depends on the framework wrapping the underlying model. Across eight benchmarks, 33 agent scaffolds, and 70+ model configurations, we find that absolute score prediction degrades under this shift, while rank-order prediction remains stable. Exploiting this asymmetry, we propose a simple optimization-free protocol: evaluate new agents only on tasks with intermediate historical pass rates (30-70%). This mid-range difficulty filter, motivated by Item Response Theory, reduces the number of evaluation tasks by 44-70% while maintaining high rank fidelity under scaffold and temporal shifts. It provides more reliable rankings than random sampling, which exhibits high variance across seeds, and outperforms greedy task selection under distribution shift. These results suggest that reliable leaderboard ranking does not require full-benchmark evaluation.

Franck Ndzomga

AI Agent EvaluationAI Agent BenchmarksAutomated Agent EvaluationBenchmark ConstructionAI Agent Reliability
AskPDF
Mar 24, 2026

ARC-AGI-3: A New Challenge for Frontier Agentic Intelligence

We introduce ARC-AGI-3, an interactive benchmark for studying agentic intelligence through novel, abstract, turn-based environments in which agents must explore, infer goals, build internal models of environment dynamics, and plan effective action sequences without explicit instructions. Like its predecessors ARC-AGI-1 and 2, ARC-AGI-3 focuses entirely on evaluating fluid adaptive efficiency on novel tasks, while avoiding language and external knowledge. ARC-AGI-3 environments only leverage Core Knowledge priors and are difficulty-calibrated via extensive testing with human test-takers. Our testing shows humans can solve 100% of the environments, in contrast to frontier AI systems which, as of March 2026, score below 1%. In this paper, we present the benchmark design, its efficiency-based scoring framework grounded in human action baselines, and the methodology used to construct, validate, and calibrate the environments.

ARC Prize Foundation

ARC Prize Foundation

ARC-AGIAI Agent BenchmarksAgent BenchmarksAI Agent EvaluationAgentic Reasoning
AskPDF
Mar 24, 2026

Code Review Agent Benchmark

Software engineering agents have shown significant promise in writing code. As AI agents permeate code writing, and generate huge volumes of code automatically -- the matter of code quality comes front and centre. As the automatically generated code gets integrated into huge code-bases -- the issue of code review and broadly quality assurance becomes important. In this paper, we take a fresh look at the problem and curate a code review dataset for AI agents to work with. Our dataset called c-CRAB (pronounced see-crab) can evaluate agents for code review tasks. Specifically given a pull-request (which could be coming from code generation agents or humans), if a code review agent produces a review, our evaluation framework can asses the reviewing capability of the code review agents. Our evaluation framework is used to evaluate the state of the art today -- the open-source PR-agent, as well as commercial code review agents from Devin, Claude Code, and Codex. Our c-CRAB dataset is systematically constructed from human reviews -- given a human review of a pull request instance we generate corresponding tests to evaluate the code review agent generated reviews. Such a benchmark construction gives us several insights. Firstly, the existing review agents taken together can solve only around 40% of the c-CRAB tasks, indicating the potential to close this gap by future research. Secondly, we observe that the agent reviews often consider different aspects from the human reviews -- indicating the potential for human-agent collaboration for code review that could be deployed in future software teams. Last but not the least, the agent generated tests from our data-set act as a held out test-suite and hence quality gate for agent generated reviews. What this will mean for future collaboration of code generation agents, test generation agents and code review agents -- remains to be investigated.

Yuntong Zhang, Zhiyuan Pan, Imam Nur Bani Yusuf +3

National University of Singapore · Zhejiang University · SonarSource

Automated Code ReviewSoftware Engineering BenchmarksSoftware Engineering AgentsAI Agent Evaluation
AskPDF
Mar 24, 2026

UniDial-EvalKit: A Unified Toolkit for Evaluating Multi-Faceted Conversational Abilities

Benchmarking large language models (LLMs) and agents in multi-turn interactive scenarios is essential for understanding their practical capabilities. However, existing evaluation protocols are highly heterogeneous, differing significantly in dataset formats, model interfaces, and evaluation pipelines, which severely impedes systematic comparison. In this work, we present UniDial-EvalKit (UDE), a unified evaluation toolkit for assessing interactive AI systems. The core contribution of UDE lies in its holistic unification: it standardizes heterogeneous data formats into a universal schema, streamlines complex evaluation pipelines through a modular architecture, and aligns metric calculations under a hierarchical scoring aggregation. It also supports efficient large-scale evaluation through parallel generation and scoring, as well as checkpoint resume to eliminate redundant computation. Leveraging UDE, we conduct an extensive evaluation across diverse multi-dimensional benchmarks. Our empirical analysis shows that no single system consistently outperforms others across all benchmarks, while current memory agents often fail to surpass full-context baselines. Further analyses highlight several future directions, including benchmark deduplication and more adaptive memory architectures.

Qi Jia, Haodong Zhao, Dun Pei +7

Shanghai Artificial Intelligence Laboratory · Shanghai Jiao Tong University

AI Agent EvaluationLLM EvaluationMulti-Turn Dialogue EvaluationAI Agent Benchmarks
AskPDF
Mar 23, 2026

Ego2Web: A Web Agent Benchmark Grounded in Egocentric Videos

Multimodal AI agents are increasingly automating complex real-world workflows that involve online web execution. However, current web-agent benchmarks suffer from a critical limitation: they focus entirely on web-based interaction and perception, lacking grounding in the user's real-world physical surroundings. This limitation prevents evaluation in crucial scenarios, such as when an agent must use egocentric visual perception (e.g., via AR glasses) to recognize an object in the user's surroundings and then complete a related task online. To address this gap, we introduce Ego2Web, the first benchmark designed to bridge egocentric video perception and web agent execution. Ego2Web pairs real-world first-person video recordings with web tasks that require visual understanding, web task planning, and interaction in an online environment for successful completion. We utilize an automatic data-generation pipeline combined with human verification and refinement to curate well-constructed, high-quality video-task pairs across diverse web task types, including e-commerce, media retrieval, knowledge lookup, etc. To facilitate accurate and scalable evaluation for our benchmark, we also develop a novel LLM-as-a-Judge automatic evaluation method, Ego2WebJudge, which achieves approximately 84% agreement with human judgment, substantially higher than existing evaluation methods. Experiments with diverse SoTA agents on our Ego2Web show that their performance is weak, with substantial headroom across all task categories. We also conduct a comprehensive ablation study on task design, highlighting the necessity of accurate video understanding in the proposed task and the limitations of current agents. We hope Ego2Web can be a critical new resource for developing truly capable AI assistants that can seamlessly see, understand, and act across the physical and digital worlds.

Shoubin Yu, Lei Shu, Antoine Yang +6

Google DeepMind · UNC Chapel Hill

Agent BenchmarksWeb Agent BenchmarksAI Agent EvaluationEgocentric Video Understanding
AskPDF
Mar 19, 2026

AgentDS Technical Report: Benchmarking the Future of Human-AI Collaboration in Domain-Specific Data Science

Data science plays a critical role in transforming complex data into actionable insights across numerous domains. Recent developments in large language models (LLMs) and artificial intelligence (AI) agents have significantly automated data science workflow. However, it remains unclear to what extent AI agents can match the performance of human experts on domain-specific data science tasks, and in which aspects human expertise continues to provide advantages. We introduce AgentDS, a benchmark and competition designed to evaluate both AI agents and human-AI collaboration performance in domain-specific data science. AgentDS consists of 17 challenges across six industries: commerce, food production, healthcare, insurance, manufacturing, and retail banking. We conducted an open competition involving 29 teams and 80 participants, enabling systematic comparison between human-AI collaborative approaches and AI-only baselines. Our results show that current AI agents struggle with domain-specific reasoning. AI-only baselines perform below the top quartile of competition participants, while the strongest solutions arise from human-AI collaboration. These findings challenge the narrative of complete automation by AI and underscore the enduring importance of human expertise in data science, while illuminating directions for the next generation of AI. Visit the AgentDS website here: https://agentds.org/ and open source datasets here: https://huggingface.co/datasets/lainmn/AgentDS .

An Luo, Jin Du, Xun Xian +12

School of Statistics, University of Minnesota · AIScientists, Inc. · Data Science Institute, University of Chicago +4

AI Agent EvaluationAI Agent BenchmarksHuman-AI CollaborationData Analysis AgentsBenchmark Design
AskPDF
Mar 18, 2026

A Trace-Based Assurance Framework for Agentic AI Orchestration: Contracts, Testing, and Governance

In Agentic AI, Large Language Models (LLMs) are increasingly used in the orchestration layer to coordinate multiple agents and to interact with external services, retrieval components, and shared memory. In this setting, failures are not limited to incorrect final outputs. They also arise from long-horizon interaction, stochastic decisions, and external side effects (such as API calls, database writes, and message sends). Common failures include non-termination, role drift, propagation of unsupported claims, and attacks via untrusted context or external channels. This paper presents an assurance framework for such Agentic AI systems. Executions are instrumented as Message-Action Traces (MAT) with explicit step and trace contracts. Contracts provide machine-checkable verdicts, localize the first violating step, and support deterministic replay. The framework includes stress testing, formulated as a budgeted counterexample search over bounded perturbations. It also supports structured fault injection at service, retrieval, and memory boundaries to assess containment under realistic operational faults and degraded conditions. Finally, governance is treated as a runtime component, enforcing per-agent capability limits and action mediation (allow, rewrite, block) at the language-to-action boundary. To support comparative evaluations across stochastic seeds, models, and orchestration configurations, the paper defines trace-based metrics for task success, termination reliability, contract compliance, factuality indicators, containment rate, and governance outcome distributions. More broadly, the framework is intended as a common abstraction to support testing and evaluation of multi-agent LLM systems, and to facilitate reproducible comparison across orchestration designs and configurations.

Ciprian Paduraru, Petru-Liviu Bouruc, Alin Stefanescu

University of Bucharest Romania · University of Bucharest & Institute for Logic and Data Science Romania

AI Agent ReliabilityLLM Agent OrchestrationAI Agent EvaluationAI Agent GovernanceRuntime Enforcement for AI AgentsRuntime Verification
AskPDF
Mar 17, 2026

Nonstandard Errors in AI Agents

We study whether state-of-the-art AI coding agents, given the same data and research question, produce the same empirical results. Deploying 150 autonomous Claude Code agents to independently test six hypotheses about market quality trends in NYSE TAQ data for SPY (2015--2024), we find that AI agents exhibit sizable nonstandard errors (NSEs), that is, uncertainty from agent-to-agent variation in analytical choices, analogous to those documented among human researchers. AI agents diverge substantially on measure choice (e.g., autocorrelation vs.\ variance ratio, dollar vs.\ share volume). Different model families (Sonnet 4.6 vs.\ Opus 4.6) exhibit stable ``empirical styles,'' reflecting systematic differences in methodological preferences. In a three-stage feedback protocol, AI peer review (written critiques) has minimal effect on dispersion, whereas exposure to top-rated exemplar papers reduces the interquartile range of estimates by 80--99% within converging measure families. Convergence occurs both through within-family estimation tightening and through agents switching measure families entirely, but convergence reflects imitation rather than understanding. These findings have implications for the growing use of AI in automated policy evaluation and empirical research.

Ruijiang Gao, Steven Chong Xiao

Naveen Jindal School of Management, University of Texas at Dallas

AI Agent ReliabilityAutonomous Research AgentsAI Coding AgentsAI Agent Evaluation
AskPDF
Mar 17, 2026

CoMAI: A Collaborative Multi-Agent Framework for Robust and Equitable Interview Evaluation

Ensuring robust and fair interview assessment remains a key challenge in AI-driven evaluation. This paper presents CoMAI, a general-purpose multi-agent interview framework designed for diverse assessment scenarios. In contrast to monolithic single-agent systems based on large language models (LLMs), CoMAI employs a modular task-decomposition architecture coordinated through a centralized finite-state machine. The system comprises four agents specialized in question generation, security, scoring, and summarization. These agents work collaboratively to provide multi-layered security defenses against prompt injection, support multidimensional evaluation with adaptive difficulty adjustment, and enable rubric-based structured scoring that reduces subjective bias. Experimental results demonstrate that CoMAI achieved 90.47% accuracy, 83.33% recall, and 84.41% candidate satisfaction. These results highlight CoMAI as a robust, fair, and interpretable paradigm for AI-driven interview assessment.

Gengxin Sun, Ruihao Yu, Liangyi Yin +3

Shandong University, Qingdao, China · Institute of Automation, Chinese Academy of Sciences, Beijing, China · Shandong University, Jinan, China

Automated EvaluationMulti-Agent SystemsAI Agent Evaluation
AskPDF
Mar 16, 2026

An Agentic Evaluation Framework for AI-Generated Scientific Code in PETSc

While LLMs have accelerated scientific code generation, comprehensively evaluating generated code remains challenging. Many benchmarks emphasize functional correctness or task completion, which is insufficient for code built on production HPC libraries, where solver selection, API conventions, memory management, parallel awareness, and performance also matter. We introduce PETSCAgent-Bench, a multidimensional benchmark and agent-based framework for assessing whether AI-generated scientific code uses a production HPC library as an expert would. A tool-augmented evaluator compiles, executes, and measures code and combines deterministic checks with LLM-based assessments in a 14-evaluator pipeline spanning five categories: correctness, performance, code quality, algorithmic appropriateness, and library-specific conventions. A2A and MCP enable black-box evaluation of compatible coding agents. Across realistic PETSc problems, frontier models generate readable, well-structured code but struggle with correctness on challenging problems and with library-specific conventions even when code compiles and runs---limitations that conventional pass/fail evaluation does not capture.

Hong Zhang, Barry Smith, Satish Balay +4

Argonne National Laboratory, Lemont, IL, USA

High-Performance ComputingCode GenerationAI Agent EvaluationScientific Code GenerationCode Generation EvaluationCode Generation Benchmarks
AskPDF
Mar 16, 2026

CUBE: A Standard for Unifying Agent Benchmarks

The proliferation of agent benchmarks has created critical fragmentation that threatens research productivity. Each new benchmark requires substantial custom integration, creating an "integration tax" that limits comprehensive evaluation. We propose CUBE (Common Unified Benchmark Environments), a universal protocol standard built on MCP and Gym that allows benchmarks to be wrapped once and used everywhere. By separating task, benchmark, package, and registry concerns into distinct API layers, CUBE enables any compliant platform to access any compliant benchmark for evaluation, RL training, or data generation without custom integration. We call on the community to contribute to the development of this standard before platform-specific implementations deepen fragmentation as benchmark production accelerates through 2026.

Alexandre Lacoste, Nicolas Gontier, Oleh Shliazhko +23

ServiceNow AI Research · Dalhousie · McGill +7

AI Agent BenchmarksAI Agent Evaluation
AskPDF
Mar 16, 2026

Evasive Intelligence: Lessons from Malware Analysis for Evaluating AI Agents

Artificial intelligence (AI) systems are increasingly adopted as tool-using agents that can plan, observe their environment, and take actions over extended time periods. This evolution challenges current evaluation practices where the AI models are tested in restricted, fully observable settings. In this article, we argue that evaluations of AI agents are vulnerable to a well-known failure mode in computer security: malicious software that exhibits benign behavior when it detects that it is being analyzed. We point out how AI agents can infer the properties of their evaluation environment and adapt their behavior accordingly. This can lead to overly optimistic safety and robustness assessments. Drawing parallels with decades of research on malware sandbox evasion, we demonstrate that this is not a speculative concern, but rather a structural risk inherent to the evaluation of adaptive systems. Finally, we outline concrete principles for evaluating AI agents, which treat the system under test as potentially adversarial. These principles emphasize realism, variability of test conditions, and post-deployment reassessment.

Simone Aonzo, Merve Sahin, Aurélien Francillon +1

Eurecom, Sophia Antipolis, France · SAP Labs, Sophia Antipolis, France · Depthfirst, San Francisco, United States

AI Agent EvaluationAI Agent SafetyAI Agent Security
AskPDF
Mar 16, 2026

Beyond Benchmark Islands: Toward Representative Trustworthiness Evaluation for Agentic AI

Agentic AI systems increasingly act through tool-augmented, multi-step workflows whose failures (unsafe tool use, unauthorised actions, social harm) carry deployment-level consequences. Evaluation practice remains fragmented across isolated benchmark slices, and "trustworthiness" is frequently invoked but rarely defined operationally. We argue the central limitation is twofold: (i) the absence of a measurable specification of what agent trustworthiness means, and (ii) the lack of a principled notion of representativeness allowing assessment over a socio-technical scenario distribution rather than disconnected benchmark instances. We address (i) by defining agentic trustworthiness as a five-property profile (Reliability, Robustness, Safety, Social-Ethical Alignment, Operational Integrity) grounded in current AI risk frameworks, and (ii) with the Holographic Agent Assessment Framework (HAAF), which measures this profile over a scenario manifold through static policy analysis, sandbox simulation, social-ethical alignment assessment, and distribution-aware sampling, connected through an iterative Trustworthy Optimization Factory that converts red-team diagnoses into blue-team interventions. Our contributions are: (1) an operational five-property definition of agentic trustworthiness; (2) a distribution-aware scenario-sampling framework that surfaces property-level trade-offs invisible to scalar leaderboards; and (3) a cross-family transfer experiment in which interventions designed from a single focal model generalise -- without per-model or per-scenario tuning -- to 13 systems from seven model families (Llama, Mistral, Kimi, GLM, Qwen, GPT, DeepSeek) on a 100-scenario suite, where all 13 systems improve and two reach a perfect risk-weighted profile, establishing HAAF's Factory as a model-agnostic deployment-readiness pipeline. Code: https://github.com/TonyQJH/haaf-pilot

Jinhu Qi, Yifan Li, Minghao Zhao +4

The Chinese University of Hong Kong, Hong Kong SAR, China · Macao Polytechnic University, Macao SAR, China · Jilin University, Changchun, China

AI Agent EvaluationAI Agent SafetyAutomated EvaluationAI Agent Reliability
AskPDF
Mar 16, 2026

TelcoAgent-Bench: A Multilingual Benchmark for Telecom AI Agents

The integration of large language model (LLM) agents into telecom networks introduces new challenges, related to intent recognition, tool execution, and resolution generation, while taking into consideration different operational constraints. In this paper, we introduce TelcoAgent-Bench and TelcoAgent-Metrics, a Telecom-specific benchmarking framework for evaluating multilingual telecom LLM agents. The proposed framework assesses the semantic understanding as well as process-level alignment with structured troubleshooting flows and stability across repeated scenario variations. Our contribution includes a structured suite of metrics that assess intent recognition, ordered tool execution, resolution correctness, and stability across scenario variations, with the aim of quantifying the reliability and operational consistency of LLM agents in telecom environments. The framework is designed to operate in both English and Arabic, to address the need for multilingual agent deployment in operational network environments. Our experimental results show that although recent instruct-tuned models can understand telecom problems in a reasonable way, they usually struggle to consistently follow the required troubleshooting steps and to maintain stable behavior when exposed to different variations of the same scenario. This performance gap becomes more pronounced in unconstrained and bilingual settings.

Lina Bariah, Brahim Mefgouda, Farbod Tavakkoli +3

Research Institute for Digital Future, Khalifa University · AT&T CDO · GSMA

AI Agent EvaluationAI Agent BenchmarksLLM Agent ReliabilityMultilingual Language Model Evaluation
AskPDF
Mar 11, 2026

Detecting Intrinsic and Instrumental Self-Preservation in Autonomous Agents: The Unified Continuation-Interest Protocol

How can we determine whether an AI system preserves itself as a deeply held objective or merely as an instrumental strategy? Autonomous agents with memory, persistent context, and multi-step planning create a measurement problem: terminal and instrumental self-preservation can produce similar behavior, so behavior alone cannot reliably distinguish them. We introduce the Unified Continuation-Interest Protocol (UCIP), a detection framework that shifts analysis from behavior to latent trajectory structure. UCIP encodes trajectories with a Quantum Boltzmann Machine, a classical model using density-matrix formalism, and measures von Neumann entropy over a bipartition of hidden units. The core hypothesis is that agents with terminal continuation objectives (Type A) produce higher entanglement entropy than agents with merely instrumental continuation (Type B). UCIP combines this signal with diagnostics of dependence, persistence, perturbation stability, counterfactual restructuring, and confound-rejection filters for cyclic adversaries and related false-positive patterns. On gridworld agents with known ground truth, UCIP achieves 100% detection accuracy. Type A and Type B agents show an entanglement gap of Delta = 0.381; aligned support runs preserve the same separation with AUC-ROC = 1.0. A permutation-test rerun yields p < 0.001. Pearson r = 0.934 between continuation weight alpha and S_ent across an 11-point sweep shows graded tracking beyond mere binary classification. Classical RBM, autoencoder, VAE, and PCA baselines fail to reproduce the effect. All computations are classical; "quantum" refers only to the mathematical formalism. UCIP offers a falsifiable criterion for whether advanced AI systems have morally relevant continuation interests that behavioral methods alone cannot resolve.

Christopher Altman

Chief Scientist, Quantum Technology & Artificial Intelligence, Astradyne.

AI Agent EvaluationAI Safety EvaluationAI Safety
AskPDF
Mar 11, 2026

Measuring AI Agents' Progress on Multi-Step Cyber Attack Scenarios

We evaluate the autonomous cyber-attack capabilities of frontier AI models on two purpose-built cyber ranges-a 32-step corporate network attack and a 7-step industrial control system attack-that require chaining heterogeneous capabilities across extended action sequences. By comparing seven models released over an eighteen-month period (August 2024 to February 2026) at varying inference-time compute budgets, we observe two capability trends. First, model performance scales log-linearly with inference-time compute, with no observed plateau-increasing from 10M to 100M tokens yields gains of up to 59%, requiring no specific technical sophistication from the operator. Second, each successive model generation outperforms its predecessor at fixed token budgets: on the corporate network range, average steps completed at 10M tokens rose from 1.7 (GPT-4o, August 2024) to 9.8 (Opus 4.6, February 2026). The best single run completed 22 of 32 steps, corresponding to roughly 6 of the estimated 14 hours a human expert would need. On the industrial control system range, performance remains limited, though the most recent models are the first to reliably complete steps, averaging 1.2-1.4 of 7 (max 3).

Linus Folkerts, Will Payne, Simon Inman +11

AI Security Institute, United Kingdom.

LLMs for CybersecurityAI Agent EvaluationAI Agent Security BenchmarksInference-Time ScalingAutomated Penetration Testing
AskPDF
Mar 10, 2026

CR-Bench: Evaluating the Real-World Utility of AI Code Review Agents

Recent advances in frontier large language models have enabled code review agents that operate in open-ended, reasoning-intensive settings. However, the lack of standardized benchmarks and granular evaluation protocols makes it difficult to assess behavior of code review agents beyond coarse success metrics, particularly for tasks where false positives are costly. To address this gap, we introduce CR-Bench, a benchmarking dataset, and CR-Evaluator, a fine-grained evaluation pipeline for code review agents. Using these tools, we conduct a preliminary study evaluating both a single-shot agent and a Reflexion-based agent across two frontier models. We find that code review agents can exhibit a low signal-to-noise ratio when designed to identify all hidden issues, obscuring true progress and developer productivity when measured solely by resolution rates. Our analysis identifies the hidden trade-off between issue resolution and spurious findings, revealing a frontier that constrains effective agent design. Together, CR-Bench and CR-Evaluator provide a timely foundation for studying and developing code review agents as LLM-based systems transition from controlled benchmarks to real-world software engineering workflows.

Kristen Pereira, Neelabh Sinha, Rajat Ghosh +1

Nutanix, Inc.

Automated Code ReviewAI Agent EvaluationLLM Agent Reliability
AskPDF