ChatGPT leads 2025 genAI usage: Cloudflare data

OpenAI’s ChatGPT remained the most used generative AI service in 2025 but faced rising competition from Anthropic’s Claude, Google’s Gemini, and Perplexity, according to Cloudflare’s Radar year-end review.

Claude rose into the number two slot on most weekdays during July and August, with Gemini overtaking Claude in mid-September.

Perplexity moved up to third place from September, while xAI’s Grok entered the top 20 in February and reached ninth by month’s end.

DeepSeek, a Chinese chatbot and open-source developer, jumped from outside the top 20 to third place in early February before stabilizing between sixth and tenth for the rest of the year.

GitHub Copilot rose to sixth among coding assistants, while ByteDance’s Doubao gained users outside China in markets including Australia, the UK, and several African countries.

ChatGPT also climbed to 33rd in Cloudflare’s overall internet services ranking by late November.

🔗 Source: The Economic Times

🧠 Food for thought

Implications, context, and why it matters.

Cloudflare’s rankings track DNS queries, not actual usage or revenue

Cloudflare Radar ranks by Domain Name System (DNS) lookups seen by the 1.1.1.1 public resolver 1, not user engagement or Application Programming Interface (API) calls or paying customers. A site that fires many DNS requests with little interaction can rank like one with heavy use.
Some enterprise (large-company) traffic may be absent when firms use internal DNS servers or network proxies (systems that route requests for users), and Claude’s stronger weekday pattern fits that enterprise focus 1.
Cloudflare uses a machine-learning-assisted ranking method from 2022 1. A lower rank can mean others grew faster, not less traffic. Moves like Perplexity at #3 or DeepSeek between #6 and #10 match DNS query trends, not market share or revenue.

Implications of Perplexity’s rise for citation/referral optimization

Perplexity’s move to third place 1 signals more use of answer engines, search-like tools that return direct answers with citations and outbound links. Cloudflare’s crawl-to-refer ratio measures how often a platform sends traffic to a site relative to how often it crawls that site 2.
Publishers can allow PerplexityBot 3 to fetch content and avoid overly restrictive robots.txt, the standard file that tells crawlers what to fetch. State attribution policies and use structured data (machine-readable markup such as Schema.org) so facts are easy to extract. See Perplexity’s crawler guidance 3.
Monitoring referrals from answer engines alongside crawl activity, for example via Cloudflare’s crawl-to-refer ratios, can reveal which pages earn citations and inform content strategy 2.