Grok Statistics (2026): 600M Monthly Users and 59 Key Stats

A data roundup of how many people use Grok, how xAI (now SpaceXSI) funds and trains it, what its models score and what they cost, taken from xAI’s own announcements, docs and pricing pages plus Pew Research Center.

xAI says its reach spans about 600 million monthly active users across the X and Grok apps (Series E announcement), but in the U.S. Grok is a niche engine: Pew found 8% of adults use it, against 44% for ChatGPT and 24% for Gemini. Grok’s numbers come from X distribution, Tesla cars, government deals and fast-moving developer models, not from mainstream search habits. For most brands, Grok is worth watching, but ChatGPT, Gemini, Perplexity and Google’s AI answers are where buyers research today, so measure those first with AI visibility tools.

Key takeaways

• 600 million monthly actives across the X and Grok apps, per xAI (Series E)
• 8% of U.S. adults use Grok, versus 44% for ChatGPT (Pew)
• $20 billion raised in xAI’s Series E before SpaceX acquired the company (Series E, SpaceX)
• One million+ H100 GPU equivalents across Colossus I and II at the end of 2025 (Series E)
• $2 / $6 per million input and output tokens for Grok 4.7 (xAI docs)
• $8 to $200 a month for consumer plans on The X Subscription (pricing)

How many people use Grok?

1. xAI’s reach spans about 600 million monthly active users across the X and Grok apps, as of January 2026 (Series E).
2. 8% of U.S. adults say they use Grok (Pew, Feb. 2026).
3. ChatGPT is used by 44% of U.S. adults, Gemini by 24%, Copilot by 17% and Meta AI by 14%, all ahead of Grok (Pew).
4. Claude is used by 6% of U.S. adults and Character.ai by 3%, both behind Grok (Pew).
5. About half of U.S. adults now use AI chatbots, up from a third in 2024 (Pew).
6. 42% of U.S. adults use chatbots to search for information, the most common use Pew measured (Pew).
7. 24% of U.S. adults use AI chatbots daily (Pew).
8. Pew’s survey covered 5,119 U.S. adults from Feb. 17 to 23, 2026 (Pew).
9. Grok Voice powers millions of conversations a day across Tesla and SpaceX (Grok Voice 3.0).
10. Grok Voice already handles tens of thousands of customer-support calls a day (Transcribe 2.0).
11. Thousands of organizations have adopted Grok Bot since launch (Grok Bot for Enterprise).
12. GenAI.mil gives 3 million military and civilian employees of the Department of War access to xAI technology (xAI for Government).
13. Grok is being deployed across more than 5,000 public schools in El Salvador, reaching over one million students (El Salvador partnership).

Grok usage in the real world

14. About 70% of Grok Voice support calls are resolved with no human in the loop, and one in four sales calls ends in a purchase (Grok Voice 3.0).
15. SpaceXSI’s support tickets rose 175% with no new hires, thanks to Grok Bot (support case study).
16. Grok Bot resolves support tickets for as little as $0.20 to $0.30 each (support case study).
17. 99% of refund requests are resolved without human intervention (support case study).
18. A procurement Grok Bot cut a $14,629 tech order to $6,143, a 58% reduction (procurement case study).

xAI funding and ownership

19. xAI raised $20 billion in its Series E, above the $15 billion it targeted (Series E).
20. NVIDIA and Cisco Investments were strategic investors in the Series E (Series E).
21. xAI raised $6 billion in its Series C in December 2024 (Series C).
22. xAI raised $6 billion in its Series B in May 2024 (Series B).
23. SpaceX acquired xAI on February 2, 2026; the company now publishes as SpaceXSI (SpaceX announcement, company timeline).
24. Cursor became part of SpaceXSI on August 14, 2026 (support case study).

Compute behind Grok

25. Colossus was fully operational in 122 days and ran workloads 19 days after the first servers arrived (Series C).
26. Grok 4’s reinforcement learning ran on Colossus, a 200,000 GPU cluster (Grok 4).
27. xAI ended 2025 with over one million H100 GPU equivalents across Colossus I and II (Series E).
28. Colossus 1 has over 220,000 NVIDIA GPUs, and SpaceXSI made it available to Anthropic in May 2026 (Anthropic compute partnership).
29. Grok 3 was trained with 10x the compute of previous state-of-the-art models (Grok 3).

Grok model stats

30. Grok-0, the first prototype, had 33 billion parameters (Announcing Grok).
31. Grok-1 is a 314 billion parameter Mixture-of-Experts model, released with open weights (Grok-1 open release).
32. 25% of Grok-1’s weights are active on a given token (Grok-1 open release).
33. Grok-1 scored 63.2% on HumanEval and 73% on MMLU (Announcing Grok).
34. Grok 3 (Think) scored 93.3% on AIME 2025 (Grok 3).
35. Grok 3 (Think) scored 84.6% on GPQA graduate-level reasoning (Grok 3).
36. Grok 3 shipped a 1 million token context window, 8 times larger than earlier Grok models (Grok 3).
37. Grok 4 Heavy was the first model to score 50.7% on Humanity’s Last Exam (text-only subset) (Grok 4).
38. Grok 4 scored 15.9% on ARC-AGI V2 (Grok 4).
39. Grok 4 Heavy scored 61.9% on USAMO 2025 (Grok 4).
40. Grok 4 Fast uses 40% fewer thinking tokens than Grok 4 for comparable scores (Grok 4 Fast).
41. Grok 4 Fast cut the price of matching Grok 4’s benchmark results by 98% (Grok 4 Fast).
42. Grok 4 Fast gave all users, including free users, the latest model (Grok 4 Fast).
43. Users preferred Grok 4.1 over the previous production model 64.78% of the time (Grok 4.1).
44. Grok 4.1 Thinking held #1 on LMArena’s Text Arena with 1483 Elo (Grok 4.1).
45. Grok 4.1 Fast has a 2-million-token context window (Grok 4.1 Fast).
46. Grok 4.5 resolved 29.0% of SWE Marathon tasks, ahead of Opus 4.8 at 26.0% (Grok 4.5).
47. Grok 4.5 scored 83.3% on Terminal Bench 2.1 (Grok 4.5).
48. Grok 4.7 scored 46.3% on CursorBench 4.0, up from 40.4% for Grok 4.6 (Grok 4.7).
49. Grok 4.7 topped LatchBio’s biosafety benchmark at 62.4% (Grok 4.7).
50. Grok 4.7 let only 3.3% of risky dual-use cyber prompts through on HackerBench (Grok 4.7).
51. Grok 4.7 has a 500k-token context window (xAI docs).
52. grok-code-fast-1 scored 70.8% on SWE-Bench-Verified (grok-code-fast-1).

Grok pricing stats

53. The Grok 4.7 API costs $2.00 per million input tokens and $6.00 per million output tokens (xAI docs).
54. xAI’s own comparison lists GPT-5.6 Sol at $4 / $20 and Fable 5.1 at $10 / $50 per million tokens (Grok 4.7).
55. grok-code-fast-1 costs $0.20 per million input tokens and $1.50 per million output tokens (grok-code-fast-1).
56. The Grok Voice Agent API bills a flat $0.05 per minute of audio (Voice Agent API).
57. xAI cut Grok-2 API prices to $2 / $10 per million tokens in December 2024 (Grok-2 1212).
58. The X Subscription’s Plus plan costs $8 a month, Premium $30 (4x Plus usage), Super $100 (15x) and Ultra $200 (50x) (pricing).
59. Premium bundles SuperGrok, Cursor Pro and X Premium+, which cost $90 a month combined before the bundle (The X Subscription).

Source: x.ai/pricing and The X Subscription, checked October 25, 2026.

What to do with this data

• Size Grok honestly. 600 million monthly actives is an X-plus-Grok figure. Pew’s 8% of U.S. adults is the better guide to how many of your buyers ask Grok for recommendations.
• Track the engines that move pipeline first. Mentionova tracks ChatGPT, Perplexity, Claude, Gemini, Google AI and Reddit, and its agents draft the content and replies that close citation gaps. It is #1 in our best AI visibility tools and LLM brand monitoring tools roundups; see Mentionova pricing for plans from $99 a month.
• Win on X if Grok matters to you. Grok reads X in real time, so active, specific posts about your category are the fastest route into its answers.
• Compare the engines. Read these numbers next to our ChatGPT statistics, Gemini statistics, Meta AI statistics, Claude statistics and AI chatbot statistics.

Frequently asked questions

How many people use Grok?

xAI said in January 2026 that its reach spans about 600 million monthly active users across the X and Grok apps. In the U.S., Pew’s February 2026 survey found 8% of adults use Grok.

Is Grok more popular than ChatGPT?

No. Pew found 44% of U.S. adults use ChatGPT and 8% use Grok, which also trails Gemini (24%), Copilot (17%) and Meta AI (14%). Grok is ahead of Claude (6%).

Who owns Grok?

Grok is built by xAI, which SpaceX acquired on February 2, 2026. The company now publishes as SpaceXSI.

How much does Grok cost?

Consumer plans on The X Subscription run from Free to Plus at $8, Premium at $30, Super at $100 and Ultra at $200 a month. The Grok 4.7 API costs $2 per million input tokens and $6 per million output tokens.

How big is the Colossus supercomputer?

xAI ended 2025 with over one million H100 GPU equivalents across Colossus I and II. Colossus 1 alone has over 220,000 NVIDIA GPUs.

How we compiled these statistics

We fetched 29 primary source pages on October 25, 2026 and recorded 62 data points in docs/content/stats-ledgers/grok-statistics.json. Every numbered figure above maps to a ledger row with a verbatim quote. User, funding, compute, model, pricing and deployment figures are xAI’s (SpaceXSI’s) own statements from its news posts, developer docs and pricing page. U.S. reach comes from Pew Research Center’s survey of 5,119 U.S. adults (Feb. 17-23, 2026). Benchmark scores are self-reported by xAI and not independently audited, and competitor figures in xAI’s charts are xAI’s own comparisons. Related reading: AI search statistics, Perplexity statistics, Microsoft Copilot statistics.