Dashboard2Code: Evaluating Multimodal Models on Reconstructing Interactive Dashboards
2026-07-06 • Software Engineering
Software EngineeringArtificial IntelligenceComputer Vision and Pattern Recognition
AI summaryⓘ
The authors introduce a new challenge called Dashboard2Code, where a model must explore and interact with a dashboard, then generate the code that recreates it. They created DashboardMimic, a dataset of 180 dashboards with different difficulty levels and common interaction types, to help test model performance. They also designed a special way to evaluate the models by checking both the code and how well the recreated dashboard works and looks, compared to the original. Their tests show that even the best models find complex dashboards hard, and open-source models lag behind closed-source ones.
data visualizationinteractive dashboardsmulti-modal large language modelsdashboard code generationPlotlyDashbenchmark datasetsemantic code analysisinteraction testingmodel evaluation
Authors
Tianhao Niu, Ziyu Han, Qiguang Chen, Shiqi Zhou, Baocai Shan, Hengjie Fang, Qingfu Zhu, Wanxiang Che
Abstract
Automatic data visualization generation has advanced rapidly with multi-modal large language models, yet existing efforts largely focus on static charts and overlook the interactive dashboards commonly used for real-world data exploration. We introduce Dashboard2Code, a novel task that requires a model to proactively explore an interactive dashboard, acquire and integrate feedback from its own interactions (e.g., clicking and filtering), and generate code that reproduces the target dashboard. To support comprehensive evaluation, we present DashboardMimic, the first Plotly+Dash benchmark for Dashboard2Code, comprising 180 carefully designed and manually verified dashboard-code pairs spanning three difficulty levels and covering eight common real-world interaction patterns. We further propose an automated evaluation framework tailored to dashboards that combines code semantic analysis with dynamic interaction-based testing to assess visual and interaction consistency, showing strong agreement with human judgments. Experiments across a range of open- and closed-source multi-modal models reveal that even the strongest systems struggle on high-complexity dashboards and that a substantial performance gap remains between open-source and closed-source models on the Dashboard2Code task.