GPT Vis

GPT-Vis is an open-source, framework-agnostic visualization library purpose-built for the LLM era. It lets AI agents and copilots generate charts using a simple, markdown-like syntax, eliminating the need for complex JSON or API calls. With 25 chart types, intelligent defaults, and streaming support, it handles incomplete or malformed data gracefully — ideal for real-time, token-by-token chart rendering. Built by AntV, GPT-Vis integrates with any JavaScript framework (React, Vue, Vanilla JS) and includes a dedicated AI agent skill for quick integration into workflows. The stable v1.0 release on May 1, 2026, marks its readiness for production use. The library's natural syntax is designed to be effortlessly generated by LLMs, making it a natural fit for AI-powered data applications. Its streaming rendering capability allows charts to appear progressively as tokens arrive, creating a smooth user experience. Fault tolerance ensures that even partially generated or malformed data doesn't break the visualization. Intelligent defaults handle data detection, color schemes, and layout automatically, reducing the need for manual configuration. Developers can get started with a single import and a few lines of code. The library includes a wrapper UI with tabs, download, and copy controls, plus support for multiple themes. The built-in isVisSyntax utility helps detect and validate streaming content. Locale support extends to UI labels for internationalization. GPT-Vis fills a unique niche for AI-generated visualizations but is not a general-purpose charting library. It complements tools like ECharts or D3 for hand-crafted charts but excels where LLM output needs to be rendered quickly with minimal configuration. For teams building AI agents, copilots, or conversational analytics, GPT-Vis offers a purpose-built, zero-cost solution.