# UI-Oceanus: Scaling GUI Agents with Synthetic Environmental Dynamics

Source: [UI-Oceanus: Scaling GUI Agents with Synthetic Environmental Dynamics](<https://www.sota2.com/research/paper/69d33e0c7d0172140ec736b6>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-07T05:00:04.607Z

Authors: Mengzhou Wu, Yuzhe Guo, Yuan Cao, Haochuan Lu, Songhe Zhu, Pingzhe Qu, Xin Chen, Kang Qin, Zhongpu Wang, Xiaode Zhang, Xinyi Wang, Wei Dai, Gang Cao, Yuetang Deng, Zhi Gong, Dezhi Ran, Linyi Li, Wei Yang, Tao Xie

Published: 2026-02-11T17:02:38.000Z

## Abstract

Scaling generalist GUI agents is hindered by the data scalability bottleneck of expensive human demonstrations and the "distillation ceiling" of synthetic teacher supervision. To transcend these limitations, we propose UI-Oceanus, a framework that shifts the learning focus from mimicking high-level trajectories to mastering interaction physics via ground-truth environmental feedback. Through a systematic investigation of self-supervised objectives, we identify that forward dynamics, defined as the generative prediction of future interface states, acts as the primary driver for scalability and significantly outweighs inverse inference. UI-Oceanus leverages this insight by converting low-cost autonomous exploration, which is verified directly by system execution, into high-density generative supervision to construct a robust internal world model. Experimental evaluations across a series of models demonstrate the decisive superiority of our approach: models utilizing Continual Pre-Training (CPT) on synthetic dynamics outperform non-CPT baselines with an average success rate improvement of 7% on offline benchmarks, which amplifies to a 16.8% gain in real-world online navigation. Furthermore, we observe that navigation performance scales with synthetic data volume. These results confirm that grounding agents in forward predictive modeling offers a superior pathway to scalable GUI automation with robust cross-domain adaptability and compositional generalization.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [GUI Action Prediction on Offline Benchmark](<https://www.sota2.com/research/sota/gui-action-prediction-on-offline-benchmark.md>): GUI Action Prediction; Offline Benchmark; Exact Match: 68.9.

- [GUI Agent Task Execution on Live mini-programs](<https://www.sota2.com/research/sota/gui-agent-task-execution-on-live-mini-programs.md>): GUI Agent Task Execution; Live mini-programs (Online); Success Rate: 44.3.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2604.02345>)

[PDF](<https://arxiv.org/pdf/2604.02345.pdf>)
