BlueLM-GUI Technical Report: A Real-Device-Centric Flywheel for Self-Improving Mobile GUI Agents
- URL: http://arxiv.org/abs/2609.12394v3
- Date: Tue, 15 Sep 2026 06:36:04 GMT
- Title: BlueLM-GUI Technical Report: A Real-Device-Centric Flywheel for Self-Improving Mobile GUI Agents
- Abstract summary: We present BlueLM-GUI, a 35B-A3B mobile GUI agent built as a real-device-centric flywheel.<n>Every Sample Matters: a dual-track pipeline with Heterogeneous Triple-System Consensus evaluation and an Error Correction & Derivation Module.<n>Every Rollout Is Real: a three-stage recipe---continual pre-training, supervised fine-tuning, and agentic reinforcement learning on hundreds of real phones.
- Score: 38.39627583261406
- License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
- Abstract: Mobile GUI agents are shifting from multi-module frameworks to native models trained end-to-end, yet industrial deployment faces three persistent gaps. Sandbox training produces a distribution mismatch with production environments; expensive real-device failures remain underutilized; and fixed benchmarks saturate, losing the power to guide iteration. We present BlueLM-GUI, a 35B-A3B mobile GUI agent built as a real-device-centric flywheel that closes these gaps through three principles. Every Sample Matters: a dual-track pipeline with Heterogeneous Triple-System Consensus evaluation and an Error Correction \& Derivation Module salvages every trajectory into usable supervision. Every Rollout Is Real: a three-stage recipe---continual pre-training, supervised fine-tuning, and agentic reinforcement learning on hundreds of real phones---grounds every rollout in real production environments, so the capability the model learns transfers directly to deployment. Every Query Evolves: a quota-driven benchmark methodology with three orthogonal axes enables precise attribution and allows the benchmark to be systematically upgraded as the model improves. BlueLM-GUI achieves 87.4 on MobileGUI-VBench, surpassing the best closed-source model by 5.1 points, and 84.9 on AndroidWorld, the best result among open-source models and competitive with closed-source models. These results demonstrate that grounding model training and iterative improvement in both real devices and the three Every principles yields strong, robust, and transferable mobile GUI capability.
Related papers
- HyMobileAgent: Data-Environment Co-Scaling for Efficient GUI Agents [57.53434680014667]
HyMobileAgent is a vision-native foundation model built on Hy3.0-VL-A3B.<n>We develop a joint data and environment centric scaling framework to address the key bottlenecks of mobile interaction.
arXiv Detail & Related papers (2026-07-16T04:12:42Z) - Xiaomi-GUI-0 Technical Report [39.05385476532886]
Xiaomi-GUI-0 is a native multimodal GUI agent for real mobile environments.<n>It is trained through a three-stage pipeline of supervised fine-tuning, step-level reinforcement learning, and agentic reinforcement learning.<n>It achieves 72.0% success on RealMobile and 78.9% on AndroidWorld, while substantially improving execution stability and abnormal-state recognition in real-world tasks.
arXiv Detail & Related papers (2026-06-30T09:36:35Z) - How Mobile World Model Guides GUI Agents? [51.29718003718467]
We train world models across four modalities: delta text, full text, diffusion-based images, and renderable code.<n>These models achieve SoTA performance on both MobileWorldBench and Code2WorldBench.<n>For overconfident mobile agents with low action entropy, posterior self-reflection provides limited gains.
arXiv Detail & Related papers (2026-05-11T10:49:31Z) - Generative Visual Code Mobile World Models [33.86938466546132]
Mobile Graphical User Interface (GUI) World Models (WMs) offer a promising path for improving mobile GUI agent performance at train- and inference-time.<n>We propose a novel paradigm: visual world modeling via renderable code generation, where a single Vision-Language Model (VLM) predicts the next GUI state as executable web code.<n>We introduce gWorld, the first open-weight visual mobile GUI WMs built on this paradigm, along with a data generation framework (gWorld) that automatically synthesizes code-based training data.
arXiv Detail & Related papers (2026-02-02T03:12:16Z) - UI-TARS-2 Technical Report: Advancing GUI Agent with Multi-Turn Reinforcement Learning [155.51875080423883]
The development of autonomous agents for graphical user interfaces presents major challenges in artificial intelligence.<n>We present UI-TARS-2, a native GUI-centered agent model that addresses these challenges through a systematic training methodology.<n> Empirical evaluation demonstrates that UI-TARS-2 achieves significant improvements over its predecessor UI-TARS-1.5.
arXiv Detail & Related papers (2025-09-02T17:44:45Z) - Mobile-Agent-v3: Fundamental Agents for GUI Automation [59.775510710011325]
This paper introduces a foundational GUI agent model that achieves state-of-the-art performance among open-source end-to-end models.<n>We propose Mobile-Agent-v3, a general-purpose GUI agent framework that further improves performance to 73.3 on AndroidWorld and 37.7 on OSWorld.
arXiv Detail & Related papers (2025-08-21T00:39:12Z) - MobileGUI-RL: Advancing Mobile GUI Agent through Reinforcement Learning in Online Environment [63.62778707277929]
MobileGUI-RL is a scalable framework that trains GUI agent in online environment.<n>It synthesizes a curriculum of learnable tasks through self-exploration and filtering.<n>It adapts GRPO to GUI navigation with trajectory-aware advantages and composite rewards.
arXiv Detail & Related papers (2025-07-08T07:07:53Z)
This list is automatically generated from the titles and abstracts of the papers in this site.
This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.