# Does ChatGPT, Claude, Perplexity or Google read llms.txt?

As of 29 September 2026, Google is the only one of these vendors to document a position, and its answer is that Google Search ignores llms.txt; OpenAI, Anthropic and Perplexity have published nothing saying ChatGPT, Claude, Perplexity or their crawlers read other sites’ llms.txt files. What is documented is narrower: some developer tools read an llms.txt that you point them at, and independent log studies find AI crawlers rarely request the file.

Updated 2026-09-29. This is the markdown version of https://llmsrank.dev/learn/does-ai-read-llms-txt.

## The short answer, vendor by vendor

“No documentation found” means we searched the vendor’s own documentation on 2026-09-29 and found no statement either way. It is not the same as “doesn’t read it”.

| Product | What the vendor documents | Date |
|---|---|---|
| **Google Search**, including AI Overviews and AI Mode | **Ignores llms.txt.** Creating one “will neither harm nor help your site’s visibility or rankings in Google Search”. | Guide added 2026-05-15; llms.txt note added 2026-06-15; page last updated 2026-07-10 |
| **Google Gemini app** | No documentation found. | Checked 2026-09-29 |
| **OpenAI** ChatGPT, GPTBot, OAI-SearchBot, ChatGPT-User | No documentation found. The crawler page covers robots.txt and doesn’t say the crawlers read llms.txt. | Page undated; checked 2026-09-29 |
| **Anthropic** Claude, ClaudeBot, Claude-User, Claude-SearchBot | No documentation found. The crawler article covers robots.txt and doesn’t mention llms.txt. | Article updated 2026-04-07; checked 2026-09-29 |
| **Perplexity** PerplexityBot, Perplexity-User | No documentation found. The crawler page mentions llms.txt only as a pointer to Perplexity’s own docs index. | Page undated; checked 2026-09-29 |
| **Microsoft** Bing, Copilot | No documentation found. Bing’s AI Performance post recommends IndexNow and doesn’t mention llms.txt. | Post of 2026-02-10; checked 2026-09-29 |
| **Chrome Lighthouse** | Audits whether the file can be fetched, in its experimental Agentic Browsing category. That is a check on your site, not a statement that any Google AI product reads the file. | Doc updated 2026-05-05 |

We found the same for Meta, Mistral and Apple: their crawler documentation, checked on 2026-09-29, covers robots.txt and doesn’t mention llms.txt.

## Google: documented, and the answer for Search is no

Google’s Search Central guide to generative AI features says it directly (page last updated 2026-07-10):

> You don’t need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search (including its generative AI capabilities), as Google Search itself doesn’t use them.
>
> Source: [Google Search Central — Optimizing your website for generative AI features on Google Search (last updated 2026-07-10)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide)

> It’s completely fine if you decide to create and maintain LLMS.txt files (or other similar files) for other services or systems that use these files. Doing so will neither harm nor help your site’s visibility or rankings in Google Search, as Google Search ignores them.
>
> Source: [Google Search Central — Optimizing your website for generative AI features on Google Search (last updated 2026-07-10)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide)

Google’s documentation changelog dates the guide to 2026-05-15 and the llms.txt note to 2026-06-15, when Google “Added a note to the AI optimization guide clarifying Google Search’s usage of llms.txt files”. Before that, Google’s position rested on remarks by its staff. The best known is John Mueller’s, reported on 2025-04-17: “AFAIK none of the AI services have said they’re using LLMs.TXT”. That is a staff remark, reported by Search Engine Journal; the guide is Google’s documented position.

The guide covers Google Search. We found no Google statement, as of 2026-09-29, about the Gemini app reading llms.txt. Google does publish one for its own Gemini API docs, at [ai.google.dev/gemini-api/docs/llms.txt](https://ai.google.dev/gemini-api/docs/llms.txt).

## OpenAI, Anthropic and Perplexity: nothing documented either way

Each of these vendors documents its crawlers and user-agent strings, and each explains how robots.txt applies. None of those pages, checked on 2026-09-29, says the crawlers or the assistant read other sites’ llms.txt files:

- **OpenAI** ([crawler page](https://developers.openai.com/api/docs/bots), undated) describes GPTBot, OAI-SearchBot and ChatGPT-User. Its only mention of llms.txt is a footer pointing to OpenAI’s own documentation index.
- **Anthropic** ([crawler article](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler), updated 2026-04-07) describes ClaudeBot, Claude-User and Claude-SearchBot, and says “Anthropic’s Bots respect ‘do not crawl’ signals by honoring industry standard directives in robots.txt”. It doesn’t mention llms.txt. The Claude API’s web fetch tool documentation doesn’t either.
- **Perplexity** ([crawler page](https://docs.perplexity.ai/guides/bots), undated) describes PerplexityBot and Perplexity-User and mentions llms.txt only as a pointer to its own docs. Several SEO pages say PerplexityBot reads llms.txt when it indexes a site; we found no Perplexity source for that claim.

> **Note: Publishing a file is not reading one.** OpenAI, Anthropic, Google (for the Gemini API), Perplexity and Mistral all publish llms.txt files for their own developer documentation, and the llms.txt proposal names the AI labs as publishers. A vendor serving its own file tells you nothing about whether its products read yours.

## What does read llms.txt: tools you point at it

The tools documented to read llms.txt read a file that a user gives them. None is documented to discover a site’s file on its own.

- **Context7 (Upstash)** lets users add a site’s llms.txt as a documentation source; an issue from 2025-05-27 shows users doing so.
- **LangChain mcpdoc** is an MCP server that gives coding tools “a user-defined list of `llms.txt` files”. Its repository was archived on 2026-09-18, so it is no longer maintained.
- **Agents with a fetch tool** can read an llms.txt when you give them its URL, as they can any text page. The Claude API’s web fetch tool, for example, is documented to fetch only URLs that have already appeared in the conversation.

Coding agents are where most of the discussion sits, but their documentation is silent too. The full Claude Code documentation, searched on 2026-09-29, never mentions llms.txt, and a user report opened on 2026-04-05 says Claude “never checks for these files by default”. As of 2026-09-29 we found no Cursor or Windsurf documentation saying either reads llms.txt on its own.

## What independent log studies found

These are third-party studies, not vendor statements. Each classifies requests by user-agent string, so they show fetches, not use: a fetch doesn’t prove the file shaped an answer.

| Study | Date | Method | Finding |
|---|---|---|---|
| [Ahrefs](https://ahrefs.com/blog/llmstxt-study/) | 2026-06-15 | 137,210 domains on Ahrefs Web Analytics, May 2026; requests classified by user agent | 28% publish a root llms.txt; “97% of those files received zero traffic in May 2026”; AI bots made 19.5% of llms.txt requests, SEO audit tools 21.7% |
| [EZY](https://www.ezy.ai/research/do-ai-bots-read-llms-txt) | 2026-07-27 | Server logs of 83 sites, 2026-04-27 to 2026-07-19; user agent plus signature checks where available | robots.txt vs llms.txt fetches: GPTBot 3,990 vs 7; ClaudeBot 3,120 vs 9; PerplexityBot 775 vs 0; Googlebot 5,125 vs 67; Meta-ExternalAgent 172 vs 193 |
| [OtterlyAI](https://otterly.ai/blog/the-llms-txt-experiment/) | 2026-02-05 | One site over 90 days, AI-oriented user agents | 84 of more than 62,100 AI bot visits went to /llms.txt (0.1%) |

A related test from [Checkly](https://www.checklyhq.com/blog/state-of-ai-agent-content-negotation/) (2026-02-19) had seven agents fetch a page with their built-in tools: Claude Code, Cursor and OpenCode asked for markdown with `Accept: text/markdown`. That is about markdown versions of pages, not llms.txt.

## What this means if you publish one

- **It is for agents and tools you, or your users, direct at it.** The proposal describes the file as read on demand, “when an agent needs information about a topic while assisting a user”. That is how the documented tools use it.
- **It is cheap to publish.** One markdown file, and Google says keeping one is “completely fine” for other services that use it.
- **It has no documented ranking effect.** Google documents that it neither helps nor harms Search. No other vendor documents any effect on visibility or citations, in either direction.
- **If you publish one, make it worth reading.** Short, described links to markdown versions of your pages serve the agents that do fetch it. See [Writing an llms.txt file](https://llmsrank.dev/learn/writing-llms-txt) and [How to check your llms.txt](https://llmsrank.dev/learn/check-llms-txt).

We will update this page when a vendor documents a position. Its date is at the top.

## Frequently asked questions

### Does ChatGPT read llms.txt?

OpenAI has not documented that it does. Its crawler page for GPTBot, OAI-SearchBot and ChatGPT-User, checked on 2026-09-29, covers robots.txt and doesn’t say the crawlers read other sites’ llms.txt files.

### Does Claude read llms.txt?

Anthropic has not documented that it does. Its crawler article (updated 2026-04-07) doesn’t mention llms.txt, and the Claude Code documentation, searched on 2026-09-29, doesn’t either. The Claude API’s web fetch tool can fetch an llms.txt whose URL you give it, like any page.

### Does Perplexity read llms.txt?

Perplexity has not documented that it does. Its crawler page, checked on 2026-09-29, mentions llms.txt only as a pointer to its own docs, and one third-party log study (EZY, 2026-07-27) recorded no PerplexityBot requests for llms.txt across 83 sites.

### Does Google use llms.txt for AI Overviews or AI Mode?

No. Google’s Search Central guide (llms.txt note added 2026-06-15, page last updated 2026-07-10) says Google Search, including its generative AI features, doesn’t use llms.txt and ignores it.

### Will an llms.txt get my site cited by AI assistants?

No vendor documents that it will. Google documents that it neither helps nor harms visibility in Search, and no other vendor has documented any effect, as of 2026-09-29.

### Is it still worth publishing an llms.txt?

It is cheap, and it serves the agents and tools that you or your users point at it, such as MCP servers and coding agents given its URL. Publish it for them, not for a ranking effect.

### Why do AI companies publish llms.txt files if they don’t document reading them?

They publish them for their own developer documentation, so that developers and their tools can feed those docs to a model. Publishing a file says nothing about whether a vendor’s products read other sites’ files.

## Sources

- [Google Search Central — Optimizing your website for generative AI features on Google Search (last updated 2026-07-10)](https://developers.google.com/search/docs/fundamentals/ai-optimization-guide)
- [Google Search Central — Latest documentation updates (llms.txt note added 2026-06-15)](https://developers.google.com/search/updates)
- [Search Engine Journal — Google says llms.txt comparable to keywords meta tag (reporting John Mueller, 2025-04-17)](https://www.searchenginejournal.com/google-says-llms-txt-comparable-to-keywords-meta-tag/544804/)
- [Chrome for Developers — Lighthouse: llms.txt audit](https://developer.chrome.com/docs/lighthouse/agentic-browsing/llms-txt)
- [OpenAI — Overview of OpenAI crawlers (GPTBot, OAI-SearchBot, ChatGPT-User)](https://developers.openai.com/api/docs/bots)
- [Anthropic — Does Anthropic crawl data from the web, and how can site owners block the crawler? (updated 2026-04-07)](https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler)
- [Claude API docs — Web fetch tool](https://platform.claude.com/docs/en/agents-and-tools/tool-use/web-fetch-tool)
- [anthropics/claude-code issue #43728 — user report (opened 2026-04-05)](https://github.com/anthropics/claude-code/issues/43728)
- [Perplexity — Perplexity crawlers (PerplexityBot, Perplexity-User)](https://docs.perplexity.ai/guides/bots)
- [Bing Webmaster Blog — Introducing AI Performance in Bing Webmaster Tools (2026-02-10)](https://blogs.bing.com/webmaster/February-2026/Introducing-AI-Performance-in-Bing-Webmaster-Tools-Public-Preview)
- [upstash/context7 issue #234 — adding llms.txt files as sources (2025-05-27)](https://github.com/upstash/context7/issues/234)
- [langchain-ai/mcpdoc — MCP server for user-defined llms.txt files (archived 2026-09-18)](https://github.com/langchain-ai/mcpdoc)
- [Ahrefs — llms.txt study, 137,210 domains (2026-06-15)](https://ahrefs.com/blog/llmstxt-study/)
- [EZY — Do AI bots read llms.txt? Server logs of 83 sites (2026-07-27)](https://www.ezy.ai/research/do-ai-bots-read-llms-txt)
- [OtterlyAI — The llms.txt experiment, one site over 90 days (2026-02-05)](https://otterly.ai/blog/the-llms-txt-experiment/)
- [Checkly — State of AI agent content negotiation (2026-02-19)](https://www.checklyhq.com/blog/state-of-ai-agent-content-negotation/)
- [llmstxt.org — The /llms.txt file, v2 (modified 2026-08-10)](https://llmstxt.org/)

## Related

- [About llms.txt](https://llmsrank.dev/learn/llms-txt.md)
- [How to check your llms.txt](https://llmsrank.dev/learn/check-llms-txt.md)
- [llms.txt vs AGENTS.md vs robots.txt](https://llmsrank.dev/learn/llms-txt-vs-agents-md.md)