GTA: Generating Long-Horizon Tasks for Web Agents at Scale

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Huang, Tenghao, Huang, Kung-Hsiang, Choubey, Prafulla Kumar, Zhou, Yilun, Chen, Muhao, May, Jonathan, Wu, Chien-Sheng
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916058227539968
author Huang, Tenghao
Huang, Kung-Hsiang
Choubey, Prafulla Kumar
Zhou, Yilun
Chen, Muhao
May, Jonathan
Wu, Chien-Sheng
author_facet Huang, Tenghao
Huang, Kung-Hsiang
Choubey, Prafulla Kumar
Zhou, Yilun
Chen, Muhao
May, Jonathan
Wu, Chien-Sheng
contents Web agents, which couple language models with browsing and tool-use capabilities, show promise as open web assistants. Yet progress is increasingly limited by the lack of scalable, process-level supervision. Existing benchmarks are largely manually constructed, providing only coarse start-goal annotations without intermediate trajectories, while recent automatic generation efforts remain expensive, biased, and shallow. These limitations prevent reliable training and evaluation of agents that must generalize to realistic, multi-hop, cross-page tasks. We introduce a scalable framework, GTA, that integrates crawling, retrieval-based seeding, in-context generation, and automated quality control to produce realistic tasks paired with executable trajectories. This design decouples crawling from generation for greater efficiency, grounds tasks in the site graph to enforce compositionality, and ensures dense supervision through deterministic replays and systematic validation. We instantiate the pipeline on over 50 websites covering e-commerce, government, forums, and news, with multilingual and multi-hop coverage. The resulting benchmark reveals a significant human-agent performance gap and enables detailed diagnostics. Our contributions are three-fold: (i) formalizing multi-hop web-agent task generation, (ii) proposing an efficient and validated pipeline for automatic data creation, and (iii) releasing a dynamic benchmark with reproducible evaluation.
format Preprint
id arxiv_https___arxiv_org_abs_2605_29218
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle GTA: Generating Long-Horizon Tasks for Web Agents at Scale
Huang, Tenghao
Huang, Kung-Hsiang
Choubey, Prafulla Kumar
Zhou, Yilun
Chen, Muhao
May, Jonathan
Wu, Chien-Sheng
Artificial Intelligence
Computation and Language
Web agents, which couple language models with browsing and tool-use capabilities, show promise as open web assistants. Yet progress is increasingly limited by the lack of scalable, process-level supervision. Existing benchmarks are largely manually constructed, providing only coarse start-goal annotations without intermediate trajectories, while recent automatic generation efforts remain expensive, biased, and shallow. These limitations prevent reliable training and evaluation of agents that must generalize to realistic, multi-hop, cross-page tasks. We introduce a scalable framework, GTA, that integrates crawling, retrieval-based seeding, in-context generation, and automated quality control to produce realistic tasks paired with executable trajectories. This design decouples crawling from generation for greater efficiency, grounds tasks in the site graph to enforce compositionality, and ensures dense supervision through deterministic replays and systematic validation. We instantiate the pipeline on over 50 websites covering e-commerce, government, forums, and news, with multilingual and multi-hop coverage. The resulting benchmark reveals a significant human-agent performance gap and enables detailed diagnostics. Our contributions are three-fold: (i) formalizing multi-hop web-agent task generation, (ii) proposing an efficient and validated pipeline for automatic data creation, and (iii) releasing a dynamic benchmark with reproducible evaluation.
title GTA: Generating Long-Horizon Tasks for Web Agents at Scale
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2605.29218