Cloudflare helped ChatGPT Search find hundreds of millions of new pages

Cloudflare and OpenAI say a small test helped ChatGPT Search find and index hundreds of millions of web pages it had never seen. The test covered about 1% of Cloudflare's network and began in July 2026. The companies shared the results on October 15. It is still a research pilot, not a product.

What changed

Cloudflare sits in front of a large share of the web. That lets it see which pages get real traffic, which have changed, and which web addresses lead nowhere. It can pass a distilled version of that picture to OpenAI's search crawler, the program that visits pages so they can be added to a search index. The inputs described so far are content freshness, traffic quality, and evidence that a page has changed.

The signals work in two directions. Some point the crawler to pages worth fetching. Others warn it away from dead addresses, sites that send almost any address to the homepage, and pages that copy other content. The companies are also testing a safety signal that could flag one page, such as a planted fake login form, without blocking the rest of a site.

The examples were everyday ones: a user guide for a washing machine, current passport requirements and fees, and a restaurant's menu and hours. Few sites may link to such pages, and crawlers that follow links tend to miss them.

What is not known

Hundreds of millions is the only count the companies gave. No exact figure has been released. The post also does not say how often the new pages were cited in ChatGPT answers, or how many visits they received. It does not name the sites in the test, and it gives no crawl-volume or server-load numbers.

Who it affects

Anyone who runs a website. A discovery signal does not grant access. A site that blocks OpenAI's search crawler stays blocked, even if Cloudflare's network flags its pages as fresh and busy. The signals also do not decide rankings. OpenAI ranks pages itself, so being indexed does not guarantee a citation in an answer.

The companies also said the pilot used web metadata for search and discovery. They said it did not share page content for model training.

• Check whether your site's settings block OpenAI's search crawler.
• Treat indexing and being cited as two separate outcomes.
• Watch for details on which sites were in the test.

What happens next

Cloudflare wants to offer the capability more widely. According to Search Engine Journal, it has signaled broad access sometime in the rest of 2026, but the October 15 announcement set no date. The next phase covers better page selection, faster and more reliable change signals, and measuring what happens after a page is found.

Two other Cloudflare moves came in the same weeks. On October 2 it opened a Web Search API in beta. In mid-October it said Verified Bots will face continuous checks, starting with whether they honor robots.txt and back off after an HTTP 429 response.