Home › Guides › IPv6 in Common Crawl
✓ checked 13 October 2026 · source: Common Crawl crawl statistics (Apache-2.0) · method · caveats · download the data
31.84% of Common Crawl’s fetches in its September 2026 crawl went over IPv6, against 0.54% in February 2026. The September crawl (CC-MAIN-2026-39) made 823.7 million IPv6 fetches out of 2.59 billion with a recorded IP version. Common Crawl says the jump came from a change to its crawler’s HTTP library, not from a change on the web.
IPv6 share, Sep 2026 crawl 31.84%HTTP/2 share, Sep 2026 78.6%crawls covered 22
Headline numbers
- A step change in March 2026. The IPv6 share went from 0.54% in CC-MAIN-2026-08 to 31.42% in CC-MAIN-2026-12, about 54 times as many IPv6 fetches. In all 7 crawls from March 2026 on, it has stayed between 31.42% and 32.49%. Before March it never passed 0.63%.
- Common Crawl’s explanation: its March 2026 announcement says an upgrade of the OkHttp library turned on Happy Eyeballs (RFC 6555), which races IPv4 and IPv6 connections and gives IPv6 a small head start. It says this raised IPv6 fetches “from 0.5% to around 31%”.
- HTTP/2 and TLS 1.3 kept climbing slowly. HTTP/2 carried 78.6% of fetches with a recorded HTTP version in September 2026, up from 72.3% in December 2024. TLS 1.3 went from 82.8% to 90.2% of TLS fetches.
IPv6 share by crawl
Context
The counters first appear in the December 2024 crawl (CC-MAIN-2024-51), when Common Crawl said its crawler could now reach IPv6-only websites but still preferred IPv4. At that point IPv6 carried 0.25% of fetches. So the figures show how often CCBot ended up on IPv6, which depends on its own connection settings as well as on how many sites support IPv6. They are not a measure of IPv6 adoption across the web.
The same statistics also count robots.txt denials: see Common Crawl Robots.txt Denials, 2016–2026.
Every crawl with IP counters
| Crawl | IPv4 fetches | IPv6 fetches | IPv6 share | HTTP/2 share | TLS 1.3 share |
|---|---|---|---|---|---|
| CC-MAIN-2026-39 | 1,763,630,004 | 823,696,081 | 31.84% | 78.6% | 90.2% |
| CC-MAIN-2026-34 | 1,720,286,489 | 828,024,206 | 32.49% | 77.5% | 89.5% |
| CC-MAIN-2026-30 | 1,738,353,307 | 821,105,985 | 32.08% | 77.5% | 89.1% |
| CC-MAIN-2026-25 | 1,678,221,484 | 806,620,998 | 32.46% | 76.5% | 89.0% |
| CC-MAIN-2026-21 | 1,752,265,702 | 807,802,165 | 31.55% | 76.6% | 88.7% |
| CC-MAIN-2026-17 | 1,754,184,699 | 834,469,633 | 32.24% | 76.2% | 88.0% |
| CC-MAIN-2026-12 | 1,604,109,903 | 734,918,582 | 31.42% | 75.3% | 86.1% |
| CC-MAIN-2026-08 | 2,532,138,475 | 13,639,533 | 0.54% | 75.2% | 86.0% |
| CC-MAIN-2026-04 | 2,697,415,922 | 14,964,642 | 0.55% | 75.4% | 85.7% |
| CC-MAIN-2025-51 | 2,524,019,254 | 13,171,938 | 0.52% | 73.6% | 85.1% |
| CC-MAIN-2025-47 | 2,662,146,105 | 15,934,055 | 0.59% | 73.7% | 85.0% |
| CC-MAIN-2025-43 | 3,005,102,966 | 18,102,799 | 0.60% | 76.3% | 86.4% |
| CC-MAIN-2025-38 | 2,781,126,217 | 17,076,506 | 0.61% | 75.6% | 86.1% |
| CC-MAIN-2025-33 | 2,823,211,683 | 18,026,023 | 0.63% | 75.1% | 85.7% |
| CC-MAIN-2025-30 | 2,817,216,601 | 17,876,010 | 0.63% | 74.4% | 85.5% |
| CC-MAIN-2025-26 | 2,794,169,955 | 15,221,146 | 0.54% | 73.9% | 85.0% |
| CC-MAIN-2025-21 | 2,891,292,644 | 13,084,143 | 0.45% | 74.3% | 84.7% |
| CC-MAIN-2025-18 | 3,214,992,172 | 10,848,876 | 0.34% | 73.5% | 84.0% |
| CC-MAIN-2025-13 | 3,203,129,525 | 10,634,370 | 0.33% | 72.9% | 84.7% |
| CC-MAIN-2025-08 | 3,128,130,576 | 10,219,007 | 0.33% | 72.8% | 83.6% |
| CC-MAIN-2025-05 | 3,491,098,648 | 8,898,076 | 0.25% | 73.8% | 83.7% |
| CC-MAIN-2024-51 | 3,057,743,967 | 7,585,839 | 0.25% | 72.3% | 82.8% |
Method
Common Crawl publishes crawler counters for every crawl in its cc-crawl-statistics repository on GitHub. On 13 October 2026 we downloaded all crawl files and kept the 22 that contain the ip_address_version counters (CC-MAIN-2024-51 to CC-MAIN-2026-39). For each crawl, the IPv6 share is the IPv6 counter divided by the sum of the IPv4 and IPv6 counters. The HTTP/2 and TLS 1.3 shares are computed the same way from the http_protocol_version and tls_protocol_version counters. A separate script recomputes every number from the raw files.
Caveats
- The shares use fetches with a recorded IP version as the denominator, not all fetch attempts. Measured against all 3.49 billion fetch attempts in September 2026, IPv6 fetches were 23.6%.
- These are counts of URL fetches, not of sites, and the crawler’s choice of URLs changes from crawl to crawl.
- The March 2026 jump reflects a crawler change, so crawls before and after it aren’t directly comparable.
- The figures describe one crawler running in one cloud network. Other clients will see different shares.
Download the data
- Common Crawl IPv6, HTTP/2 and TLS 1.3 by crawl, 2024–2026 (CSV): every crawl with IPv4, IPv6, HTTP/2 and TLS 1.3 counts and shares.
Cite as: ProxyPicker, “IPv6 in Common Crawl, 2024–2026”, 13 October 2026, https://proxypicker.com/common-crawl-ipv6/. Data: Common Crawl crawl statistics.
Related: Common Crawl Robots.txt Denials · AI-Crawler & llms.txt Census.
Press & data requests: email [email protected] for interviews, the raw data behind this page, or corrections. See also our contact page.