Agentic Benchmarks

Recent advances in world models have shown promise for modeling future dynamics of environmental states, enabling agents to reason and act without accessing real environments. Current methods mainly perform single-step or fixed-horizon rollouts, leaving their potential for complex task planning under-exploited. We propose Imagine-then-Plan (\texttt{ITP}), a unified framework for agent learning via lookahead imagination, where an agent's policy model interacts with the learned world model, yielding multi-step ``imagined'' trajectories. Since the imagination horizon may vary by tasks and stages, we introduce a novel adaptive lookahead mechanism by trading off the ultimate goal and task progress. The resulting imagined trajectories provide rich signals about future consequences, such as achieved progress and potential conflicts, which are fused with current observations, formulating a partially \textit{observable} and \textit{imaginable} Markov decision process to guide policy learning. We instantiate \texttt{ITP} with both training-free and reinforcement-trained variants. Extensive experiments across representative agent benchmarks demonstrate that \texttt{ITP} significantly outperforms competitive baselines. Further analyses validate that our adaptive lookahead largely enhances agents' reasoning capability, providing valuable insights into addressing broader, complex tasks. Our code and data will be publicly available at https://github.com/loyiv/ITP.

Multi-agent systems based on large language models, particularly centralized architectures, have recently shown strong potential for complex and knowledge-intensive tasks. However, central agents often suffer from unstable long-horizon collaboration due to the lack of memory management, leading to context bloat, error accumulation, and poor cross-task generalization. To address both task-level memory inefficiency and the inability to reuse coordination experience, we propose StackPlanner, a hierarchical multi-agent framework with explicit memory control. StackPlanner addresses these challenges by decoupling high-level coordination from subtask execution with active task-level memory control, and by learning to retrieve and exploit reusable coordination experience via structured experience memory and reinforcement learning. Experiments on multiple deep-search and agent system benchmarks demonstrate the effectiveness of our approach in enabling reliable long-horizon multi-agent collaboration.

Large Language Models (LLMs) are increasingly serving as autonomous agents, and their utilization of external tools via the Model Context Protocol (MCP) is considered a future trend. Current MCP evaluation sets suffer from issues such as reliance on external MCP services and a lack of difficulty awareness. To address these limitations, we propose MCPAgentBench, a benchmark based on real-world MCP definitions designed to evaluate the tool-use capabilities of agents. We construct a dataset containing authentic tasks and simulated MCP tools. The evaluation employs a dynamic sandbox environment that presents agents with candidate tool lists containing distractors, thereby testing their tool selection and discrimination abilities. Furthermore, we introduce comprehensive metrics to measure both task completion rates and execution efficiency. Experiments conducted on state-of-the-art LLMs reveal significant performance differences in handling complex, multi-step tool invocations. All code is open-source at https://github.com/Brunestuder/MCPAgentBench.

Climate science demands automated workflows to transform comprehensive questions into data-driven statements across massive, heterogeneous datasets. However, generic LLM agents and static scripting pipelines lack climate-specific context and flexibility and thus perform poorly in practice. We present ClimateAgent, an autonomous multi-agent framework that orchestrates end-to-end climate data analytic workflows. ClimateAgent decomposes user questions into executable subtasks coordinated by an Orchestrate-Agent and a Plan-Agent; acquires data via specialized Data-Agents that dynamically introspect APIs to synthesize robust download scripts; and completes analysis and reporting with a Coding-Agent that generates Python code, visualizations, and a final report with a built-in self-correction loop. To enable systematic evaluation, we introduce Climate-Agent-Bench-85, a benchmark of 85 real-world tasks spanning atmospheric rivers, drought, extreme precipitation, heat waves, sea surface temperature, and tropical cyclones. On Climate-Agent-Bench-85, ClimateAgent achieves 100% task completion and a report quality score of 8.32, outperforming GitHub Copilot (6.27) and a GPT-5 baseline (3.26). These results demonstrate that our multi-agent orchestration with dynamic API awareness and self-correcting execution substantially advances reliable, end-to-end automation for climate science analytic tasks. The source code of ClimateAgent is available at https://github.com/Relaxed-System-Lab/ClimateAgent.

Multimodal Large Language Models (MLLMs) show promising results as decision-making engines for embodied agents operating in complex, physical environments. However, existing benchmarks often prioritize high-level planning or spatial reasoning, leaving the fine-grained action intelligence required for embodied physical interaction underexplored. To address this gap, we introduce CFG-Bench, a new benchmark designed to systematically evaluate this crucial capability. CFG-Bench consists of 1,368 curated videos paired with 19,562 question-answer pairs spanning three evaluation paradigms targeting four cognitive abilities: 1) Physical Interaction, 2) Temporal-Causal Relation, 3) Intentional Understanding, and 4) Evaluative Judgment. Together, these dimensions provide a systematic framework for assessing a model's ability to translate visual observations into actionable knowledge, moving beyond mere surface-level recognition. Our comprehensive evaluation on CFG-Bench reveals that leading MLLMs struggle to produce detailed instructions for physical interactions and exhibit profound limitations in the higher-order reasoning of intention and evaluation. Moreover, supervised fine-tuning (SFT) on our data demonstrates that teaching an MLLMs to articulate fine-grained actions directly translates to significant performance gains on established embodied benchmarks. Our analysis highlights these limitations and offers insights for developing more capable and grounded embodied agents. Project page: https://cfg-bench.github.io/

Simulating human reasoning in open-ended tasks has long been a central aspiration in AI and cognitive science. While large language models now approximate human responses at scale, they remain tuned to population-level consensus, often erasing the individuality of reasoning styles and belief trajectories. To advance the vision of more human-like reasoning in machines, we introduce HugAgent (HUman-Grounded AGENT Benchmark), which rethinks human reasoning simulation along three dimensions: (i) from averaged to individualized reasoning, (ii) from behavioral mimicry to cognitive alignment, and (iii) from vignette-based to open-ended data. The benchmark evaluates whether a model can predict a specific person's behavioral responses and the underlying reasoning dynamics in out-of-distribution scenarios, given partial evidence of their prior views. HugAgent combines structured questionnaires with semi-structured think-aloud interviews to collect ecologically valid belief states, belief updates, and reasoning traces from human participants. Our experiments reveal a clear asymmetry: models recover a person's belief state from their own context reasonably well, but struggle to predict belief updates under intervention. Cross-person and cross-domain controls trace this gap to associative matching within a topic rather than identity-consistent reasoning, suggesting that progress requires better-calibrated change detection, not simply more context. We scope the benchmark to self-reported belief reasoning in three policy domains: healthcare, surveillance, and zoning. The benchmark, along with its complete data collection pipeline and companion chatbot, is open-sourced as HugAgent (https://github.com/jajamoa/HugAgent) and TraceYourThinking (https://github.com/jajamoa/trace-your-thinking).

Large Language Model (LLM) agents are powering a growing share of interactive web applications, yet remain vulnerable to misuse and harm. Prior jailbreak research has largely focused on single-turn prompts, whereas real harassment often unfolds over multi-turn interactions. In this work, we present the Online Harassment Agentic Benchmark consisting of: (i) a synthetic multi-turn harassment conversation dataset, (ii) a multi-agent (e.g., harasser, victim) simulation informed by repeated game theory, (iii) three jailbreak methods attacking agents across memory, planning, and fine-tuning, and (iv) a mixed-methods evaluation framework. We utilize two prominent LLMs, LLaMA-3.1-8B-Instruct (open-source) and Gemini-2.0-flash (closed-source). Our results show that jailbreak tuning makes harassment nearly guaranteed with an attack success rate of 95.78--96.89% vs. 57.25--64.19% without tuning in Llama, and 99.33% vs. 98.46% without tuning in Gemini, while sharply reducing refusal rate to 1-2% in both models. The most prevalent toxic behaviors are Insult with 84.9--87.8% vs. 44.2--50.8% without tuning, and Flaming with 81.2--85.1% vs. 31.5--38.8% without tuning, indicating weaker guardrails compared to sensitive categories such as sexual or racial harassment. Qualitative evaluation further reveals that attacked agents reproduce human-like aggression profiles, such as Machiavellian/psychopathic patterns under planning, and narcissistic tendencies with memory. Counterintuitively, closed-source and open-source models exhibit distinct escalation trajectories across turns, with closed-source models showing significant vulnerability. Overall, our findings show that multi-turn and theory-grounded attacks not only succeed at high rates but also mimic human-like harassment dynamics, motivating the development of robust safety guardrails to ultimately keep online platforms safe and responsible.

Today's AI models learn primarily through mimicry and refining, so it is not surprising that they struggle to solve problems beyond the limits set by existing data. To solve novel problems, agents should acquire skills by exploring and learning through experience. Finding a scalable learning mechanism for developing agents that learn through interaction remains a major open problem. In this work, we introduce BuilderBench, a benchmark to accelerate research into agent training that centers open-ended exploration. BuilderBench requires agents to learn how to build any structure using blocks. BuilderBench is equipped with (1) a simulator of a robot interacting with various physical blocks, and (2) a task-suite with over 50 diverse target structures that are carefully curated to test an understanding of physics, mathematics, and long-horizon planning. Agents are provided with a target structure at the start, and can interact with the environment for multiple episodes to experiment and learn various skills for building the structure. Solving these tasks requires \emph{embodied reasoning} in a way that is not reflected in words but rather in actions, experimenting with different strategies and piecing them together. Our experiments with multiple state-of-the-art frontier language model based agents and tabula rasa reinforcement learning algorithms show that these agents cannot solve any of the non-trivial tasks in the BuilderBench. Our analysis throws light on the lack of exploration abilities in these models.

Autonomous agents navigating human society must master both production activities and social interactions, yet existing benchmarks rarely evaluate these skills simultaneously. To bridge this gap, we introduce StarDojo, a novel benchmark based on Stardew Valley, designed to assess AI agents in open-ended production-living simulations. In StarDojo, agents are tasked to perform essential livelihood activities such as farming and crafting, while simultaneously engaging in social interactions to establish relationships within a vibrant community. StarDojo features 1,000 meticulously curated tasks across five key domains: farming, crafting, exploration, combat, and social interactions. Additionally, we provide a compact subset of 100 representative tasks for efficient model evaluation. The benchmark offers a unified, user-friendly interface that eliminates the need for keyboard and mouse control, supports all major operating systems, and enables the parallel execution of multiple environment instances, making it particularly well-suited for evaluating the most capable foundation agents, powered by multimodal large language models (MLLMs). Extensive evaluations of state-of-the-art MLLMs agents demonstrate substantial limitations, with the best-performing model, GPT-4.1, achieving only a 12.7% success rate, primarily due to challenges in visual understanding, multimodal reasoning and low-level manipulation. As a user-friendly environment and benchmark, StarDojo aims to facilitate further research towards robust, open-ended agents in complex production-living environments.

AI agents have the potential to aid users on a variety of consequential tasks, including conducting scientific research. To spur the development of useful agents, we need benchmarks that are challenging, but more crucially, directly correspond to real-world tasks of interest. This paper introduces such a benchmark, designed to measure the accuracy of AI agents in tackling a crucial yet surprisingly challenging aspect of scientific research: computational reproducibility. This task, fundamental to the scientific process, involves reproducing the results of a study using the provided code and data. We introduce CORE-Bench (Computational Reproducibility Agent Benchmark), a benchmark consisting of 270 tasks based on 90 scientific papers across three disciplines (computer science, social science, and medicine). Tasks in CORE-Bench consist of three difficulty levels and include both language-only and vision-language tasks. We provide an evaluation system to measure the accuracy of agents in a fast and parallelizable way, saving days of evaluation time for each run compared to a sequential implementation. We evaluated two baseline agents: the general-purpose AutoGPT and a task-specific agent called CORE-Agent. We tested both variants using two underlying language models: GPT-4o and GPT-4o-mini. The best agent achieved an accuracy of 21% on the hardest task, showing the vast scope for improvement in automating routine scientific tasks. Having agents that can reproduce existing work is a necessary step towards building agents that can conduct novel research and could verify and improve the performance of other research agents. We hope that CORE-Bench can improve the state of reproducibility and spur the development of future research agents.

Agent skill libraries are becoming routable software assets: a retrieved skill can contribute instructions, scripts, resource bindings, and execution assumptions to an agent. This makes retrieval failures more specific than broad irrelevance. A system can find the right capability family yet expose the wrong same-capability representative. We study this failure as same-capability risk-exposure retrieval. Each benchmark unit pairs a helpful skill with a query-specific risky sibling that shares the capability family but differs on an execution-controlling contract, such as the required resource, precondition, procedure, or artifact. We introduce SameCapRisk-Bench, an auditable benchmark with 1,190 skill-risk units and 1,686 evaluation query cases: 694 marked-sibling units under public library pressure and 496 hard role-flip units where the same two skills swap helpful/risky roles across paired queries. The release records admission evidence, cue/leakage checks, source hashes, family relations, and fixed candidate pools. The benchmark reports helpful ranking together with harmful sibling rate (HSR@K), the top-K exposure of the marked risky sibling. On this benchmark, public SkillRouter, SkillRet, and R3-Skill retrieve helpful skills at high Recall@3 (0.848--0.888) but also expose marked risky siblings frequently (HSR@3 0.346--0.372). A fully public score-and-cluster pipeline lowers HSR@3 to 0.128--0.182, with Recall@3 of 0.713--0.776. Under a benchmark-trained reference scorer, public text-cluster and controlled resolvers reach HSR@3 0.012 and 0.007; the latter attains Recall@3 0.833. Skill retrieval should therefore report both capability matching and same-family risk exposure, with HSR serving as a targeted exposure certificate for fixed skill libraries.

Evaluating agents on the growing number of agentic benchmarks is challenging because they often require complex environments and agent integrations. We introduce Harbor Adapters, a unified evaluation infrastructure for agentic benchmarks. Our work makes three contributions. First, we develop benchmark adapters that port more than 80 benchmarks to evaluate arbitrary agents, and validate them through rigorous code review and parity experiments. Second, we conduct a large-scale evaluation of 8 models spanning capability tiers across 54 benchmarks; every model is run with Terminus-2 and with one of 3 native harnesses. This enables a broader analysis of agent capabilities and failure modes than was previously possible. Third, we introduce Harbor-Index, a curated set of 82 difficult, diverse, and high-quality tasks spanning 29 benchmarks, refined from the adapted suite through difficulty filtering, AI and human audit, and an audit-and-fix loop. Harbor-Index preserves the challenge and breadth of large-scale agentic evaluations while being affordable to run; no evaluated model-harness configuration exceeds 30% pass rate, and the strongest (GPT-5.5 with Codex) reaches 28.0%. We release the adapters, evaluation results, in-depth analysis, and Harbor-Index as open-source artifacts to support more reliable and comprehensive evaluation of language-model agents.

Deep Research Agents (DRA) aim to help users search the web, synthesize information, and deliver comprehensive investigative reports. Prior benchmarks often either under-evaluate a system's ability to produce meaningful insights and high-quality writing, or adopt coarse or LLM-defined criteria that are hard to verify and can diverge from human expert judgment. To address these issues, we introduce Deep Research Bench II, a new benchmark for evaluating DRAs. It contains 132 grounded research tasks across 22 domains; for each task, an agent must produce a research report that is evaluated by a set of 9,430 fine-grained binary rubrics in total, covering three dimensions: information recall, analysis, and presentation. All rubrics are derived from carefully selected expert-written investigative articles and are constructed through a four-stage LLM+human pipeline that combines automatic extraction with over 400 human-hours of expert review, ensuring that the criteria are verifiable and aligned with human expert judgment. We evaluate several state-of-the-art deep-research agents on Deep Research Bench II and find that even the strongest models satisfy fewer than 50% of the rubrics, revealing a substantial gap between current DRAs and human experts. We release the benchmark, evaluation scripts, and all rubrics at https://github.com/imlrz/DeepResearch-Bench-II to facilitate future research on deep-rearch agents.