diff --git a/_quarto.yml b/_quarto.yml
index 10a0f35ec..e807e5e70 100644
--- a/_quarto.yml
+++ b/_quarto.yml
@@ -315,6 +315,7 @@ website:
                 - docs/websites/website-drafts.qmd
                 - docs/websites/website-search.qmd
                 - docs/websites/website-tools.qmd
+                - docs/websites/website-llms.qmd
                 - docs/websites/website-about.qmd
                 - section: "Listing Pages"
                   href: docs/websites/website-listings.qmd
diff --git a/docs/prerelease/1.9/_highlights.qmd b/docs/prerelease/1.9/_highlights.qmd
index 90ed2761b..8e2e6c327 100644
--- a/docs/prerelease/1.9/_highlights.qmd
+++ b/docs/prerelease/1.9/_highlights.qmd
@@ -2,6 +2,8 @@ Quarto 1.9 includes the following new features:
 
 - [Posit Connect Cloud Publishing](/docs/publishing/posit-connect-cloud.qmd): Publish documents and websites to Posit Connect Cloud using `quarto publish posit-connect-cloud`.
 
+- [Output for LLMs](/docs/websites/website-llms.qmd): Generate `llms.txt` and `.llms.md` files for websites, making your content more accessible to large language models and AI-powered tools.
+
 - [Privacy-first cookie consent](/docs/websites/website-tools.qmd#cookie-consent): The default for cookie consent has changed to `type: express`, providing opt-in consent that blocks cookies until users explicitly agree. This privacy-conscious default is designed with modern privacy regulations in mind.
 
 - [`aria-label` for videos](/docs/authoring/videos.qmd#accessibility-label): Improve accessibility of embedded videos by providing custom descriptive labels for screen readers instead of the default "Video Player" label.
diff --git a/docs/websites/website-llms.qmd b/docs/websites/website-llms.qmd
new file mode 100644
index 000000000..9589642b3
--- /dev/null
+++ b/docs/websites/website-llms.qmd
@@ -0,0 +1,61 @@
+---
+title: Output for LLMs
+---
+
+## Overview
+
+Quarto can generate [llms.txt](https://llmstxt.org/) format output for your website, making your content more accessible to large language models and AI-powered tools. When enabled, Quarto creates an `llms.txt` index file and `.llms.md` markdown files for each page.
+
+## Enabling `llms.txt`
+
+To enable `llms.txt` output, add `llms-txt: true` to your website configuration:
+
+```{.yaml filename="_quarto.yml"}
+website:
+  title: "My Documentation"
+  llms-txt: true
+```
+
+## Generated Output
+
+When you render your site, Quarto creates files in the output directory (typically `_site/`).
+
+### `llms.txt` Index
+
+An `llms.txt` file is created at the root of your site. This index lists all pages with their titles, using your site's title and description from the website configuration in `_quarto.yml`.
+
+### Markdown Files
+
+For each HTML page, a corresponding `.llms.md` file is created alongside it (e.g., `guide.html` gets `guide.llms.md`). These contain clean markdown versions of your content.
+
+## Content Conversion
+
+HTML content is converted to markdown using Pandoc. Navigation elements, sidebars, breadcrumbs, footers, and scripts are removed. Tables, code blocks, callouts, tabsets, and images are converted to standard markdown formats. Internal links are automatically rewritten from `.html` to `.llms.md` extensions, including links with anchors.
+
+## Conditional Content
+
+To include content only in the `.llms.md` output, use the `llms-txt` format with [conditional content](/docs/authoring/conditional.qmd) classes.
+
+For example, to include content only in `document.llms.md`:
+
+```{.markdown filename="document.qmd"}
+::: {.content-visible when-format="llms-txt"}
+This content appears only in the `.llms.md` output.
+:::
+```
+
+To exclude content from `document.llms.md`:
+
+```{.markdown filename="document.qmd"}
+::: {.content-hidden when-format="llms-txt"}
+This content appears in HTML but not in `.llms.md` output.
+:::
+```
+
+## Draft Handling
+
+Draft pages are automatically excluded from both `.llms.md` generation and the `llms.txt` index. The 404 page is also excluded from the index.
+
+## Build Behavior
+
+On full builds, all `.llms.md` files and the `llms.txt` index are regenerated. On incremental builds, individual `.llms.md` files are updated but `llms.txt` is preserved, matching the behavior of sitemap generation.