Family
Benchmark · 2026
Dashboard2Code
A benchmark for reconstructing interactive Plotly Dash applications from screenshots, optional DOM, and interaction.
Open the primary source ↗Task
Rebuild 180 dashboards across 450 tasks and reproduce their visible and callback behavior.
Evidence
Visual, code, dynamic-behavior, and callback checks calibrated against human ratings.
What it establishes
Shows that DOM access improves exploration and that static screenshots hide factually wrong interactive state.
What it does not establish
The fixed desktop, single-framework setting omits mobile, animation, popups, assistive paths, and reader outcomes.
Related entities
Continue through the library.
- DashArenaA benchmark, browser executor, and human-calibrated judge for open-ended interactive dashboard generation.
- RealChart2CodeA chart-to-code benchmark built from real source data, multi-panel tasks, and iterative refinement.
- Data Formulator 2A visualization authoring system that combines natural-language requests with direct controls, visible data transformations, and history.