titleGTAGym: AAA Game Tasks Environment for Evaluation and Reinforcement Learning
authorsJiajun Chen, Shicheng Fang, Jiahao Lu, Yuxin Wang, Xipeng Qiu
date2026-11-11
venuearXiv (preprint, not peer reviewed)
idarXiv:2611.13028
affiliationOpenMOSS
upvotes11 (Hugging Face Daily Papers, 2026-11-14)
linksabs pdf hf code
cite
relatedMOSS-Image Technical Report
World Action Models: The Next Frontier in Embodied AI
SWE-bench Science: Can Coding Agents Resolve Engineering Tasks in Science?
MOSS-VL Technical Report
In-Context World Modeling for Robotic Control
abstractAAA games offer a rich testbed for evaluating agent perception, reasoning, and control. However, evaluating whether agents can complete tasks in AAA games is challenging and often human-annotated because task outcomes are often difficult to verify automatically, let alone applying it for Reinforcement Learning (RL). We introduce GTAGym, an environment and framework for evaluation and RL on GTA V tasks. Its evaluation benchmark, GTAVBench, derived from GTA V main-story missions, contains 103 online tasks reconstructed across nine task families, alongside 47 image-based shooting and sniping tasks using GTA V imagery. A web server provides access to the online environment and evaluation interface, supporting parallel testing through screenshots and keyboard-mouse actions with automatic engine-state outcome checks. We evaluate six frontier LLMs on GTAVBench with textual task instructions and on its GTAVBench-AGI branch, where instructions appear only in the first screenshot and a harness automatically compresses interaction history. The best configuration achieves 70.9% overall accuracy in GTAVBench-AGI, yet aerial navigation, evasion and interception remain difficult. GTAGym also provides a scalable generator that samples task parameters within user-specified ranges and option sets under fixed templates, creating batches of 3D and image-based shooting environments for RL with automatic reward and termination signals. LoRA RL of Qwen3.8-27B on generated on-foot navigation tasks improves the in-domain performance and transfers to out-of-domain task categories, and native GTA V tasks from 15% to 23%.

bibtex

@misc{chen2026gtagym,
  title={GTAGym: AAA Game Tasks Environment for Evaluation and Reinforcement Learning},
  author={Jiajun Chen and Shicheng Fang and Jiahao Lu and Yuxin Wang and Xipeng Qiu},
  year={2026},
  eprint={2611.13028},
  archivePrefix={arXiv},
  url={https://arxiv.org/abs/2611.13028}
}