Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2610.20424

BizBench: Unveiling Hidden Challenges in Deploying Long-Horizon Agents for Real-World Business

Enhan Zhao·Yongqi Tong·Xuejun Yu·Hang Wang·Pan Wang·Yichao Wang·Di He·Jianshe Li·Wei Wu
View original source ↗Read source document ↗Export metadata ↓

Abstract

Existing long-horizon agent benchmarks often emphasize final task success while providing limited assessment of intermediate execution validity. Yet real-world business operations require agents to navigate extensive, potentially outdated or conflicting policies while satisfying critical constraints throughout execution. These requirements are particularly consequential in high-stakes domains such as payments, where intermediate violations can incur financial losses or regulatory consequences regardless of the final outcome. To bridge this evaluation gap, we introduce BizBench, a multi-turn benchmark grounded in real-world procedures from a leading payments company. Its executable environment integrates 789 knowledge documents and 601 discoverable business tools, with consistent policies, states, and tool behavior. The benchmark comprises 50 tasks involving resource constraints, temporal dependencies, prior commitments, and conflicting or outdated guidance. Each task is validated through reference execution, while deterministic checks assess both goal attainment and execution validity without prescribing a single solution path. Experiments reveal substantial limitations even among frontier models: Claude Opus 5 and GPT-6 Sol achieve strict success rates of only 64.0% and 61.3%, respectively. For Qwen3.8-27B, goal attainment reaches 42.0%, but strict success drops to 32.7%, demonstrating how outcome-based evaluation can mask invalid execution. Beyond success rates, DeepSeek-V4.1-Flash averages 191.3 tool calls per run, approximately 70% more than reference executions, revealing substantial execution overhead. Overall, BizBench establishes a testbed for evaluating and advancing long-horizon agents beyond task completion, toward reliable and constraint-compliant execution in complex business environments.

Explore related subjects

cs.CE

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Enhan Zhao, Yongqi Tong, Xuejun Yu, Hang Wang, Pan Wang, Yichao Wang, Di He, Jianshe Li, Wei Wu. 2026-10-14. BizBench: Unveiling Hidden Challenges in Deploying Long-Horizon Agents for Real-World Business. https://arxiv.org/abs/2610.20424

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

BioXArena: Benchmarking LLM Agents on Multi-Modal Biomedical Machine Learning Tasks

Large language model (LLM) agents increasingly automate machine-learning model building, yet biomedical benchmarks emphasize question answering, reasoning, and tool use or cover narrow slices of biomedical ML coding. We introduce BioXArena, a biomedical machine learning (BioML) coding benchmark for task-specific model building on heterogeneous, often multi-modal datasets. It contains 76 end-to-end tasks across 9 domains: sequence, single-cell, structure, network biology, chemical biology, perturbation dynamics, phenotype--disease, imaging, and text-integrated tasks. Each task packages primary-source data with hidden labels, held-out graders, and biology-aware metrics on a unified 0-to-100 scale; agents write runnable code, train models, and predict private test samples. BioXArena emphasizes realistic data interfaces: most tasks require agents to integrate multiple input sources, including 46 tasks that combine distinct modality families. We compare 18 harness--backbone configurations: general-purpose coding, biomedical, and ML agents plus our Minimal runner, all under the same controlled resource configuration. Codex with GPT-6 Astra achieves the highest task-level average score of 73.91 among 18 configurations, closely followed by Claude Code with Opus 5.5 (73.79). Matched-backbone comparisons show higher observed scores for mature coding systems than for Minimal; seven Minimal backends characterize model differences under a fixed runner. Beyond the main leaderboard, ablations, robustness checks, and scaling, runtime, and failure-mode analyses characterize how backbones, scaffolds, budgets, and domains affect BioML coding performance, exposing modeling and execution gaps across agent and LLM backbone families.

cs.CE↗
arXiv2026

Stochastic Latent Twins: Learning Stochastic Solution Operators

Scientific prediction under uncertainty requires more than a single future state: it requires the distribution of possible outcomes. We show that this distribution-valued evolution can be learned directly as a stochastic solution operator between arbitrary source and target times. We introduce Stochastic Latent Twins, which represent the physical state through a deterministic encoder-decoder while placing stochasticity entirely in a probability-valued latent evolution map. This separation makes the stochastic transition law independent of the state representation and enables direct probabilistic prediction without recursive time stepping. We establish an end-to-end Wasserstein approximation theory that separates representation and latent-transition errors, derive stability guarantees, and develop a distributional theory of physics conformity with latent-to-physical transfer bounds and conformity by construction. The resulting framework can be learned directly from observed transitions and supports forward prediction, backward conditional evolution, one-shot generation, and physics-conforming stochastic dynamics within the same formulation. Numerical experiments ranging from linear and non-Gaussian SDEs to stochastic Lorenz dynamics, generative modeling, a stochastic Hamiltonian system, and a two-dimensional stochastic Burgers equation verify the theory, expose the distinct roles of state representation and transition-family expressivity, compare favorably with recent stochastic and one-step generative methods, and demonstrate exact physical conservation by construction. More broadly, Stochastic Latent Twins provide a step toward scientific surrogates that represent not only the evolution of a system, but also its uncertainty and physical structure, opening new opportunities across uncertainty quantification, data assimilation, inverse problems, control, and digital twins.

cs.CE↗
arXiv2026

Product Costing in Data Pipelines

Large data-intensive enterprises spend millions of dollars monthly running data pipelines. Apportioning this cost back to individual products, known as product costing, is a prerequisite for data pricing and cost recovery. However, this process is fundamentally more challenging than physical manufacturing costing due to the unique characteristics of data pipelines, including the non-rivalrous nature of data that complicates the costing formulation, the prevalence of cyclic dependencies that distort the accuracy of results, and the massive scale of pipelines that leads to computational intractability. In this paper, for the first time, we formalize the problem of product costing in data pipelines as a manufacturing cost of data products (MCDP) problem to overcome the non-rivalrous nature of data. To solve MCDP, we propose CostApp, a hybrid algorithm that achieves both exactness and scalability, even for cyclic and massive-scale pipelines. CostApp performs a dimensionality reduction by isolating cyclic dependencies via a feedback arc set (FAS). It aggregates costs onto a compact set of key vertices using a scalable iterative approach, transforming the intractable computational problem into a reduced linear system that can be solved exactly and efficiently by a basic matrix-based method. Experiments on both real-world data and synthetic data show that CostApp scales to process a production pipeline with over 28 million entities within an hour while preserving exactness, supporting pipelines up to 10^2 larger than the direct matrix-based baseline.

cs.CE↗