Fast, Interactive Python Charting Library · XY<br>For AI agents: the complete XY documentation index is at llms.txt. Markdown versions are available by appending .md or sending Accept: text/markdown.<br>NewXY's initial launch is here. Get startedGet started
Learn
Gallery
API Reference
Learning<br>Overview
What is xy?
Installation
Your First Chart
Benchmarks
Core Concepts
Composition Model
Data and Columns
Axes and Scales
Interactions and Selections
Large Data and Performance
Configuration
Styling
Overview
Examples
Customize Each Part
Chrome Slots
Animations
Themes and Export
Capability Matrix
Advanced Styling Gallery
Advanced
XY Architecture
Custom Marks
Runtime and Deployment
Charts<br>Overview
Core Charts
Line
Area, Step & Stairs
Scatter
Bar and Column
Distributions
Histogram
Box Plot
Violin Plot
ECDF
Density & Fields
Heatmap
Hexbin
Contour
Specialized
Uncertainty
Stem
Segments
Sankey
Funnel
Polar Charts
Overview
Radar
Radial Bar
Pie & Donut
Wind Rose
Components
Overview
Marks
Axes
Legends
Tooltips
Colorbars
Modebars & Controls
Annotations
Triangle Mesh
Facets and Layers
Other<br>Reflex
Notebooks
Matplotlib (xy.pyplot)
Guides
Overview
DataFrames and Real Data
Display and Export
Real-time and Streaming Data
Dashboards and Linked Views
Serving, CSP, and Offline Use
Deployment Recipes
Troubleshooting
Getting Help
Reference
Overview
Chart Factories
Marks and Components
Figure Methods
Events and Callbacks
Public Types
Limitations and Alpha Status
Changelog
Contributing
Fast, Interactive Python Charting Library
div]:!p-0">What is xy?<br>XY is a Python charting library for interactive 2D visualizations that stay<br>smooth at millions of points and take your design system seriously. Two ideas<br>shape the library:<br>Fast, even with lots of data. XY draws the detail you can actually see<br>instead of every row at once, so pan, zoom, and hover stay responsive as<br>your data grows.<br>Styled by your CSS, not ours. Titles, axes, legends, tooltips, and controls<br>are addressable with plain CSS or Tailwind through 23 stable slots, and your<br>design tokens reach the marks themselves — in the browser, in SVG, and in<br>native PNG. What each mechanism reaches is published per renderer in the<br>Capability Matrix, including where it<br>stops.
All four interactive charts are live — drag to pan, scroll to zoom, and hover<br>to inspect exact values. Together they render more than a million points from a<br>single probability field across four chart families.<br>View the customizable Python source.
Early alpha. XY is pre-1.0. The declarative composition model is stabilizing, but callback<br>payloads, the Reflex adapter, chart breadth, and adaptive-rendering thresholds<br>may change. See Limitations and alpha status<br>before committing to a long-lived integration.
Start here<br>Browse the visual gallery to see the chart<br>families available today.<br>Install XY and build<br>your first chart.<br>Learn the composition model behind every chart.<br>Read the benchmark snapshot with its output<br>contracts and measurement caveats.<br>Follow the styling overview for CSS, Tailwind, theme<br>tokens, and rendered-mark styles.<br>Why XY<br>Python teams usually face a trade-off: charting libraries that hit an<br>interactivity ceiling as data grows, or browser-first tools that give up design<br>control. XY is built for the workflows where that trade-off bites. Compose<br>marks, axes, legends, and controls in Python; brand them with CSS, Tailwind,<br>and theme tokens; and ship the same chart to notebooks, applications, and<br>standalone HTML, PNG, or SVG exports.<br>Performance is part of the architecture, not an option flag. Native Rust<br>kernels aggregate data before display, binary transport keeps numbers out of<br>JSON, and the WebGL2 client bounds browser work by what the screen can show,<br>while exact source data stays in Python for hover and selection.<br>The numbers back this up. In the recorded live interactive sweep, XY reached a<br>correct, stable canvas in 0.071 seconds at 10,000 points and 0.081 seconds at<br>100 million. Matplotlib reached 13.4 seconds at 50 million points, while Plotly<br>reached 9.8 seconds at 25 million.<br>Live interactive render time<br>Correct and stable canvas · Apple M5 Pro · lower is better
XY<br>XY · density off<br>Matplotlib<br>Plotly
Above 200,000 rows, XY's default path switches to a density view while keeping<br>exact source rows available for deeper zooms. With density disabled, the same<br>engine still rendered 100 million individual markers in 1.34 seconds. The<br>benchmark publishes both paths so the effect of aggregation stays visible.<br>Inspect the benchmark evidence or<br>browse the chart gallery.<br>Install it and see for yourself:
Browse the chart gallery or jump straight to<br>your first chart.
Built with Reflex