OpenAI Launches ChatGPT Sites for Building Web Apps Without Code

OpenAI has introduced ChatGPT Sites, a new capability that allows users to create and publish interactive websites and lightweight applications using conversational prompts, according to documentation published by the company.

The feature, now in public beta, enables subscribers to build web experiences by simply describing what they want in natural language. Users can request dashboards, project trackers, launch calendars, prototypes, internal portals, and reports without writing code themselves.

How ChatGPT Sites works

To create a site, users access ChatGPT Work mode on the web or desktop app and describe their desired website, optionally mentioning @Sites to trigger the feature. They can include content, files, data, links, and constraints in their prompt. ChatGPT then generates a preview that users can refine through follow-up requests before deploying.

Each deployment generates a production URL immediately. Users who want to review changes without updating a live site must save a version first. Custom domains are supported where available, though users must own the domain and manage DNS records themselves.

Editing follows the same conversational pattern. Users can request changes to copy, layout, data, styles, links, forms, or interactive behavior, review the updated preview, and publish when satisfied.

Access controls and availability

ChatGPT Sites is available on Pro, Pro Lite, Plus, Business, Enterprise, and Edu plans, but not on Free or Go tiers. The feature is not available in the European Economic Area, Switzerland, or the United Kingdom at launch. OpenAI is rolling out access gradually, with Pro and Enterprise users receiving it first, followed by Plus and Business subscribers.

New sites default to private access, visible only to the owner and workspace administrators. Depending on plan and workspace settings, users can share with selected individuals or groups, anyone in their workspace, or publish publicly to the internet. In Enterprise workspaces, public publishing is disabled by default and requires administrator approval.

Why it matters

ChatGPT Sites represents a significant expansion of generative AI beyond text and images into full application deployment. By eliminating traditional development workflows, OpenAI is positioning ChatGPT as an end-to-end platform for creating functional web experiences, not just generating code snippets. This could accelerate internal tool development for businesses while raising questions about security, data governance, and quality control when non-technical users can deploy public-facing applications with minimal oversight.

Usage limits and restrictions

During the public beta, usage is subject to plan-specific limits that apply across all sites on an account. Users approaching limits receive notifications and may be unable to create new sites or maintain high-usage sites publicly until usage decreases.

The feature includes important restrictions: sites cannot process protected health information or payment card data, target children under 13, enable financial transactions, distribute malware, or violate OpenAI's usage policies. Data residency and inference residency are not supported at launch.

For Business and Enterprise customers, OpenAI does not use conversations or site information to train models by default. For Free, Go, Plus, and Pro users, conversations may be used for training if the "Improve the model for everyone" setting is enabled.

The details were first reported in OpenAI's help documentation.