Claude statistics (2026)

This page collects 97 Claude and Anthropic statistics. Company figures come from Anthropic's own announcements, research reports and pricing pages; usage estimates come from Sensor Tower, Similarweb and Pew, and market share from Menlo Ventures, each labelled with its year. The first section is Rank.ai's own data on how Claude answers real buyer questions compared with Gemini and ChatGPT.

Rank.ai data: how Claude answers buyer questions

We tracked 11,726 fresh answers from ChatGPT (gpt-4o-mini), Claude (claude-haiku-4-5) and Gemini (gemini-2.5-flash) with web search on, collected from May 7 to October 16, 2026, for 102 buyer questions tracked by 15 Rank.ai customer accounts, excluding cached repeats. Engines were queried through their APIs with Rank.ai's own web search tool. Compare the other two engines on our Gemini statistics and ChatGPT statistics pages.

• 2,577distinct domains cited by Claude across 20,931 citations, the widest range of the three (ChatGPT 2,448, Gemini 1,230).Rank.ai analysis of 11,726 AI answers, May to October 20262026
• 5.40citations per Claude answer on average, the fewest of the three (ChatGPT 6.34, Gemini 7.38).Rank.ai analysis of 11,726 AI answers, May to October 20262026
• 80.1%of Claude answers include at least one citation, against 94.4% for Gemini and 96.3% for ChatGPT.Rank.ai analysis of 11,726 AI answers, May to October 20262026
• 98.4%of Claude answers ran a web search before answering.Rank.ai analysis of 11,726 AI answers, May to October 20262026
• 36.6%of Claude answers name at least one business, the lowest of the three (Gemini 38.1%, ChatGPT 39.0%).Rank.ai analysis of 11,726 AI answers, May to October 20262026
• 22.2%of Claude answers name five or more businesses, the highest share, while 63.4% name none.Rank.ai analysis of 11,726 AI answers, May to October 20262026
• 2.47businesses named per Claude answer on average (Gemini 2.53, ChatGPT 2.03).Rank.ai analysis of 11,726 AI answers, May to October 20262026
• 8.1%of Claude answers name the brand that tracks the question, and 4.5% cite that brand's own site.Rank.ai analysis of 11,726 AI answers, May to October 20262026
• 1,706characters is the average Claude answer (Gemini 1,666, ChatGPT 2,139).Rank.ai analysis of 11,726 AI answers, May to October 20262026
• 3.62%of Claude's citations go to Reddit, 2.22% to YouTube and 0.04% to Wikipedia.Rank.ai analysis of 11,726 AI answers, May to October 20262026
• 0%of Claude's local plumber picks lacked a matching Google Maps listing, against 5.3% for Perplexity, 6.6% for Gemini and 7.7% for ChatGPT, in 60 answers with 460 picks across 4 engines and 5 cities.loc.ai local AI index, October 20262026
• 1,097Google reviews is the median for Claude's local picks, the highest of the four engines (ChatGPT 713).loc.ai local AI index, October 20262026

Claude users and traffic

Anthropic does not publish a monthly user count, so the newest figures are third-party estimates from app and web panels, labelled with their source.

• 9.6%of visits to AI chatbot websites went to Claude in August 2026, up from 2% a year earlier, while ChatGPT fell from 78% to 57%.Similarweb, Anthropic IPO analysis2026
• 824 millionmonthly visits to claude.ai in April 2026, up from 203 million in February.Similarweb, Anthropic IPO analysis2026
• 44.8%six-month retention for people who started using Claude in February 2026, against 25.8% for the January 2025 group and 59% for ChatGPT.Similarweb, Anthropic IPO analysis2026
• 245 millionestimated monthly Claude users in May 2026.TechCrunch, reporting Sensor Tower data2026
• 10.3%of the global AI assistant audience used Claude at the end of May 2026.TechCrunch, reporting Sensor Tower data2026
• 13%of Anthropic's users pay for a subscription, Sensor Tower estimates.TechCrunch, reporting Sensor Tower data2026
• 452%year-over-year growth in Claude's audience in May 2026.Sensor Tower, State of AI 20262026
• 14%US market share for Claude by May 2026, up from 4.4% a year earlier.Sensor Tower, State of AI 20262026
• 120minutes a month spent per Claude user, up 40 minutes.Sensor Tower, State of AI 20262026
• 5days in March 2026 (March 1 to 5) when Claude had more daily downloads than ChatGPT.Sensor Tower, State of AI 20262026
• 2x+growth in Claude's consumer paid subscriptions from the start of 2026 to late March, per an Anthropic spokesperson.TechCrunch, Claude popularity with paying consumers2026
• 6%of US adults have ever used Claude, including 10% of those aged 18 to 29 and 1% of those 65 and over.Pew Research Center, chatbots Americans use2026

Revenue, run-rate and funding

Anthropic reports run-rate revenue, its current monthly revenue multiplied by 12, with each funding round. It went from about $1 billion to over $47 billion in 16 months.

• $65 billionrun-rate revenue at the end of July 2026, shared with investors and reported by Bloomberg; Anthropic declined to comment.Bloomberg via BGov, Anthropic revenue run rate2026
• $47 billion+run-rate revenue in early May 2026.Anthropic, Series H announcement2026
• $965 billionpost-money valuation in the $65 billion Series H on May 28, 2026, which included $5 billion from Amazon.Anthropic, Series H announcement2026
• $30 billion+run-rate revenue on April 6, 2026, up from about $9 billion at the end of 2025.Anthropic, Google and Broadcom compute partnership2026
• $14 billionrun-rate revenue on February 12, 2026, after growing over 10x a year in each of the three prior years.Anthropic, Series G announcement2026
• $380 billionpost-money valuation in the $30 billion Series G, led by GIC and Coatue.Anthropic, Series G announcement2026
• $100 billion+committed by Anthropic to AWS over 10 years, as Amazon invested $5 billion more with up to $20 billion to follow.Anthropic, Amazon compute agreement2026
• $5 billion+run-rate revenue in August 2025, up from about $1 billion at the start of 2025.Anthropic, Series F announcement2025
• $183 billionpost-money valuation in the $13 billion Series F led by ICONIQ, September 2025.Anthropic, Series F announcement2025
• $10 billionin investment commitments from NVIDIA and up to $5 billion from Microsoft, with Anthropic buying $30 billion of Azure compute.Anthropic, Microsoft and NVIDIA partnerships2025
• $61.5 billionpost-money valuation in the $3.5 billion Series E led by Lightspeed, March 2025.Anthropic, Series E announcement2025
• $8 billiontotal committed by Amazon by November 2024, after an initial $4 billion in September 2023.Anthropic, Amazon and Trainium2024
• $124 millionSeries A, led by Jaan Tallinn, in May 2021.Anthropic, Series A announcement2021

Business adoption

Most of Anthropic's revenue comes from businesses using Claude through its API, cloud partners and team plans. Customers spending over $1 million a year doubled in under two months in early 2026.

• 46,000partner firms with 175,000 Claude certifications, plus a $100 million commitment to train 10,000 deployed engineers by the end of 2027.Anthropic, Claude Frontier Academy2026
• 1,500+GovTech Singapore engineers and technical staff use Claude Code.Anthropic, Singapore office opening2026
• 100,000+customers run Claude on Amazon Bedrock.Anthropic, Amazon compute agreement2026
• 1,000+customers spend over $1 million a year, as of April 6, 2026.Anthropic, Google and Broadcom compute partnership2026
• 500+customers spent over $1 million a year in February 2026, up from about a dozen two years earlier.Anthropic, Series G announcement2026
• 8of the Fortune 10 are Claude customers.Anthropic, Series G announcement2026
• 300,000+business customers in September 2025.Anthropic, Series F announcement2025
• 7xgrowth in one year in large accounts, those above $100,000 in run-rate revenue.Anthropic, Series F announcement2025
• $37 billionspent by enterprises on generative AI in 2025, up from $11.5 billion in 2024.Menlo Ventures, State of Generative AI in the Enterprise2025

Claude Code

Claude Code is Anthropic's coding agent, which works in a developer's terminal. It reached $1 billion in run-rate revenue six months after general availability.

• $2.5 billion+Claude Code run-rate revenue in February 2026.Anthropic, Series G announcement2026
• 2xgrowth in Claude Code weekly active users from January 1 to February 12, 2026, while business subscriptions quadrupled.Anthropic, Series G announcement2026
• 50%+of Claude Code revenue came from enterprises in February 2026.Anthropic, Series G announcement2026
• 4%of public GitHub commits were authored by Claude Code in early 2026, with SemiAnalysis projecting over 20% by year end.SemiAnalysis, Claude Code is the inflection point2026
• $1 billionClaude Code run-rate revenue in November 2025, when Anthropic acquired Bun.Anthropic, Claude Code reaches $1B milestone2025
• $500 million+Claude Code run-rate revenue in September 2025, with usage up over 10x in three months.Anthropic, Series F announcement2025
• 79%of Claude Code conversations were automation, where Claude does the task, against 49% on Claude.ai.Anthropic, Impact on software development2025
• 59%of their work is done with Claude, Anthropic engineers reported, with a self-reported 50% productivity gain, in a survey of 132 staff.Anthropic, How AI is transforming work at Anthropic2025
• 116%rise in consecutive tool calls per Claude Code transcript, from 9.8 to 21.2, between February and August 2025.Anthropic, How AI is transforming work at Anthropic2025
• February 24, 2025: Claude Code launched as a research preview with Claude 3.7 Sonnet, and became generally available in May 2025.Anthropic, Claude 3.7 Sonnet and Claude Code2025

Enterprise market share

Menlo Ventures, an Anthropic investor, surveys enterprise buyers on which model providers they pay for. Anthropic passed OpenAI in enterprise LLM spend in 2025.

• 40%of enterprise LLM spend went to Anthropic in 2025, up from 24% in 2024 and 12% in 2023.Menlo Ventures, State of Generative AI in the Enterprise2025
• 54%of enterprise spend on coding models went to Anthropic in late 2025, against 21% for OpenAI.Menlo Ventures, State of Generative AI in the Enterprise2025
• 27%of enterprise LLM spend went to OpenAI in 2025, down from 50% in 2023.Menlo Ventures, State of Generative AI in the Enterprise2025
• 32%of enterprise LLM API usage was Anthropic in mid-2025, against 25% for OpenAI and 20% for Google.Menlo Ventures, 2025 Mid-Year LLM Market Update2025
• 42%of the code generation market was Anthropic in mid-2025, against 21% for OpenAI.Menlo Ventures, 2025 Mid-Year LLM Market Update2025
• $8.4 billionof enterprise model API spend in mid-2025, up from $3.5 billion in November 2024.Menlo Ventures, 2025 Mid-Year LLM Market Update2025

Model releases, benchmarks and pricing

Anthropic ships Claude in sizes: Haiku (small), Sonnet (mid), Opus (large) and, since June 2026, Fable and Mythos. Benchmark scores are Anthropic's own, from each launch post; prices are per million input and output tokens.

• 1 milliontoken context window and 128,000-token maximum output on all four current models: Fable 5.1, Opus 5.5, Sonnet 5.5 and Haiku 5.5.Anthropic Docs, Models overview2026
• $0.10per million input tokens and $0.50 per million output tokens for Claude Haiku 5.5 (prompts up to 100,000 tokens), launched October 7, 2026.Anthropic, Claude Haiku 5.52026
• 70.6%on Terminal-Bench 4.0 for Claude Sonnet 5.5, launched September 28, 2026 at $2 and $10.Anthropic, Claude Sonnet 5.52026
• 81.8%on OSWorld 2.1 for Claude Opus 5.5, launched September 22, 2026 at $4 and $20, which Anthropic says costs 40% less to run than Opus 5.Anthropic, Claude Opus 5.52026
• 60.9%on Terminal-Bench 4.0 for Claude Mythos 5.1, and 55.8% for Fable 5.1, in September 2026.Anthropic, Claude Fable and Mythos 5.12026
• 3xthe next-best model's ARC-AGI 3 score for Claude Opus 5, launched July 24, 2026 at $5 and $25.Anthropic, Claude Opus 52026
• $10per million input tokens and $50 per million output for Claude Fable 5 and Mythos 5, launched June 9, 2026.Anthropic, Claude Fable 5 and Mythos 52026
• 81.42%on SWE-bench Verified for Claude Opus 4.6, February 5, 2026, the first Opus model with a 1 million token context (in beta).Anthropic, Claude Opus 4.62026
• $20a month for Claude Pro, from $100 for Max, and $20 per seat for Enterprise; the Batch API is 50% off.Claude, Pricing2026
• $5per million input tokens and $25 per million output for Claude Opus 4.5, November 24, 2025, a two-thirds cut from Opus 4's $15 and $75.Anthropic, Claude Opus 4.52025
• 73.3%on SWE-bench Verified for Claude Haiku 4.5, October 15, 2025, at $1 and $5.Anthropic, Claude Haiku 4.52025
• 77.2%on SWE-bench Verified and 61.4% on OSWorld for Claude Sonnet 4.5, September 29, 2025, which worked for over 30 hours on multi-step tasks.Anthropic, Claude Sonnet 4.52025
• 74.5%on SWE-bench Verified for Claude Opus 4.1, August 5, 2025.Anthropic, Claude Opus 4.12025
• 72.5%on SWE-bench Verified for Claude Opus 4 and 72.7% for Sonnet 4, May 22, 2025.Anthropic, Introducing Claude 42025
• 63.7%on SWE-bench Verified for Claude 3.7 Sonnet, February 24, 2025.Anthropic, Claude 3.7 Sonnet2025
• 64%on Anthropic's agentic coding eval for Claude 3.5 Sonnet in June 2024, against 38% for Claude 3 Opus, at twice the speed.Anthropic, Claude 3.5 Sonnet2024
• 200,000token context window for the Claude 3 family, launched March 4, 2024.Anthropic, Claude 3 family2024

How people use Claude

The Anthropic Economic Index studies anonymized Claude conversations to see which tasks people bring to it. Coding leads, and the share of fully delegated tasks has grown.

• 35%of Claude.ai conversations on weekdays are personal, rising to just under 50% on weekends, from a survey of about 9,700 users.Anthropic, Economic Index June 2026 report2026
• 35%of Claude.ai conversations were computer and math tasks in February 2026.Anthropic, Economic Index March 2026 report2026
• 49%of jobs have had at least a quarter of their tasks done with Claude.Anthropic, Economic Index March 2026 report2026
• 52%of Claude.ai conversations were augmentation, where people work with Claude, in November 2025, against 45% automation.Anthropic, Economic Index: new building blocks2026
• 39%of Claude.ai conversations were fully delegated in August 2025, up from 27% in late 2024.Anthropic, Economic Index September 2025 report2025
• 77%of first-party API transcripts showed automation patterns in August 2025.Anthropic, Economic Index September 2025 report2025
• 7.0is Israel's per-capita Claude usage index, the highest, against 4.57 for Singapore, 3.62 for the US and 0.27 for India (1.0 matches population share).Anthropic, Economic Index September 2025 report2025
• 37.2%of conversations were computer and math tasks in the first Index, from about 1 million conversations.Anthropic, The Anthropic Economic Index2025
• 36%of jobs used Claude for at least a quarter of their tasks in early 2025.Anthropic, The Anthropic Economic Index2025

Safety, compute and infrastructure

Anthropic publishes safety results alongside its models and has signed compute deals with Google, Amazon and Microsoft.

• 181working Firefox exploits developed by Claude Mythos Preview, against 2 for Opus 4.6; over 99% of the flaws found were unpatched at publication.Anthropic, Mythos Preview research2026
• $100 millionin usage credits and $4 million in donations committed through Project Glasswing to secure open-source software.Anthropic, Project Glasswing2026
• 1,000+hours of external bug-bounty testing on Fable 5 found no universal jailbreak; safeguards trigger in under 5% of sessions.Anthropic, Claude Fable 5 and Mythos 52026
• 5 GWof new AWS capacity for Anthropic, with over 1 million Trainium2 chips in use.Anthropic, Amazon compute agreement2026
• 1 millionGoogle TPUs available to Anthropic under an expansion worth tens of billions of dollars, with well over a gigawatt online in 2026.Anthropic, Expanding use of Google Cloud TPUs2025
• $50 billioninvested in US data centers with Fluidstack in Texas and New York, creating about 800 permanent jobs.Anthropic, $50 billion in American AI infrastructure2025
• 100+security controls activated with ASL-3 protections for Claude Opus 4 in May 2025.Anthropic, Activating ASL-3 protections2025
• 4.4%jailbreak success rate with Constitutional Classifiers, against 86% without, after 183 red-teamers spent over 3,000 hours.Anthropic, Constitutional Classifiers2025
• $25,000top bug bounty for a universal jailbreak in 2025, up from $15,000 in 2024.Anthropic, New bug bounty program2025

What this means for marketers

• Claude is a business buyer's assistant. Most of its revenue comes from companies, so B2B brands should check how it describes them. See AI search statistics and GEO and AEO statistics.
• A wide citation pool helps smaller sites. Claude drew on twice as many domains as Gemini in our sample, so niche pages have a better shot at being cited. Our guide to AI SEO tools covers how to track it.
• For local businesses, review volume shows up. Claude's local picks had a median of 1,097 Google reviews; see getting Google reviews.
• Check each engine. Claude, Gemini and Perplexity answer the same question differently. The free AI grade shows whether they name your business today.

Sources and method

Company figures come from Anthropic's announcements, research reports, model pages, docs and pricing page, each opened and checked on 2026-10-16. Benchmarks are Anthropic's own published scores. User and traffic estimates come from Sensor Tower (via TechCrunch), Similarweb and Pew Research Center, and market share from Menlo Ventures, an Anthropic investor. One revenue figure is Bloomberg's report and is labelled as such. Rank.ai's figures come from 11,726 fresh answers from ChatGPT (gpt-4o-mini), Claude (claude-haiku-4-5) and Gemini (gemini-2.5-flash) with web search on, for 102 buyer questions tracked by 15 customer accounts, collected May 7 to October 16, 2026, queried through each model's API with Rank.ai's web search tool, cached repeats excluded. Local figures come from the loc.ai local AI index, October 2026. We left out user counts, demographics and download figures we could only find in roundups.