Home / Reports / 21 November 2026

Data report

AI tools search interest in the US, 21 October–21 November 2026

Snapshot taken on 21 November 2026 at 15:34 CET, Europe/Amsterdam. Written by Cleanrows.

Snapshot of Google Trends taken on 21 November 2026 at 15:34 CET. Data collected with Google Trends API – Bulk Keywords, Compare & Trending Now by Cleanrows. Every number below comes from two real runs the same afternoon (16 billed keyword-rows, about $0.005 / 0.11 CZK all-in). These are relative search-interest figures, not product usage, installs or revenue.

What we measured

Eight AI-tool search terms, web search, United States, time range today 1-m (21 October to 21 November 2026):

ChatGPT, Google Gemini, Claude AI, Microsoft Copilot, Perplexity AI, Grok, Midjourney, Cursor AI.

The comparison uses one shared 0–100 scale, so the numbers can be read against each other. A second run pulled each term on its own 0–100 scale, only to list related queries. Those per-term averages are not comparable across tools, and we do not report them as a ranking.

Query spelling matters. We used the more specific forms (Google Gemini, Claude AI, Microsoft Copilot, Perplexity AI, Cursor AI) to cut down on other meanings. Grok is the short form people actually type; related queries for it are mostly the xAI product (grok ai, grok xai, grok bot), with a little crypto-prediction noise. We did not measure the even shorter names Gemini, Claude, Copilot, Perplexity or Cursor on their own. The last day in the series, 21 November, is marked partial by Google. Weekends are visibly quieter than weekdays for every term, which is normal for work-related searches — do not read the partial Saturday value as a crash.

Key numbers

  • ChatGPT dominates the set. On the shared scale its average interest is 83.4. The next term, Google Gemini, averages 7.6 — about eleven times smaller. ChatGPT peaked at 100 on 21 October and was still at 83 on 20 November, the last complete day.
  • Google Gemini is the only term that is clearly rising (change_percent +15.7, direction rising). It peaked at 10 on 16 November and held 10 on 17–19 November.
  • Claude AI and Grok are tied on average interest at 5.0. Claude is slightly up over the window (+6.0, stable); Grok is slightly down (−10.8, falling). Claude had its highest complete-day value, 6.09, on 17 November.
  • Everyone else sits close to zero on this scale, because ChatGPT is so large. Microsoft Copilot averages 0.8, Perplexity AI 0.4, Midjourney 0.2, Cursor AI 0.1. Copilot and Cursor AI are flagged precision: "low": they are hard to measure next to a term that peaks at 100.
  • Copilot's spike on 12 November looks like an outage, not a product surge. On 12 November Copilot reached its peak of 2 on the shared scale (100 on its own scale). Its rising related queries are almost all outage searches: "microsoft copilot outage", "is microsoft copilot down right now", and several close variants, all marked Breakout.
  • Midjourney is falling, and its rising queries are mostly rivals. Average 0.2, change_percent −24.1. Breakout related searches include "nano banana 2", "midjourney vs nano banana", "lovart", "novelai", "google veo" and "best image generating ai", as well as "midjourney valuation".

change_percent compares the average of the last quarter of the timeline with the first quarter. trend_direction is the Actor's label (rising / falling / stable) from that figure.

Ranking on one shared 0–100 scale (US, last month)

RankQueryAverage interestPeak (date)20 Nov (last complete day)Change %DirectionPrecision
1ChatGPT83.39100 (21 Oct)83−2.21stablehigh
2Google Gemini7.6110 (16 Nov)8+15.69risinghigh
3Claude AI5.036.09 (17 Nov)5.30+5.98stablehigh
4Grok5.036 (21 Oct)4−10.81fallinghigh
5Microsoft Copilot0.812 (12 Nov)10.00stablelow
6Perplexity AI0.420.55 (21 Oct)0.37−19.81fallinghigh
7Midjourney0.220.30 (22 Oct)0.18−24.14fallinghigh
8Cursor AI0.130.18 (21 Oct)0.12−29.41fallinglow

Claude AI and Grok share the same average (5.03); Claude is ranked third because its peak was slightly higher. Several mid-sized terms show a finer decimal because the comparison is chained through ChatGPT as the anchor. Latest-day values for 21 November are omitted here because that day was still partial when the run finished.

What people searched next (related queries)

These lists come from the second run, where each keyword is scaled on its own. They are useful as clues, not as proof that a product shipped.

  • ChatGPT. Top related: "chatgpt login", "chatgpt free", "chatgpt app", "chatgpt business". Rising: "danish man sues openai chatgpt" (+1,250%) and "chatgpt 官网" (+50%). No product-launch Breakouts in this window.
  • Google Gemini. Breakout related searches included "google gemini 4 carbon model" and "google gemini 4 hydrogen features", plus "google gemini personal intelligence tips" (+700%) and "google home smart speaker with gemini" (+140%). These are searches Google labelled rising, not a confirmation of a launch.
  • Claude AI. Top related already includes "claude code" (value 66). Breakout: "claude fable 5.2". Also rising: "claude ai deletes william baude files" (+2,650%), which looks like a news event.
  • Microsoft Copilot. Top related is Microsoft 365 Copilot. Rising is dominated by downtime queries (see above), which lines up with the 12 November peak.
  • Perplexity AI. No rising queries came back. Top related is mostly "what is perplexity ai" and comparisons with ChatGPT, Claude and Gemini.
  • Grok. Top related: "grok bot", "grok ai", "what is grok", "grok xai". Breakouts included "elon musk grok ai odyssey", "spacex grok bot usage limits" and "grok ai xrp price prediction" — the last of those is a reminder that a short name still attracts noise.
  • Midjourney. Rising queries clustered around other image generators, listed above.
  • Cursor AI. Top related is still explanatory: "what is cursor ai", "cursor ai coding", "cursor ai pricing". Breakout: "cursor ai which company". Also rising: "loveable ai" and "google antigravity".

How to reproduce it

import os
from apify_client import ApifyClient

client = ApifyClient(os.environ["APIFY_TOKEN"])

KEYWORDS = [
    "ChatGPT", "Google Gemini", "Claude AI", "Microsoft Copilot",
    "Perplexity AI", "Grok", "Midjourney", "Cursor AI",
]

# Shared 0-100 scale for the ranking
compare = client.actor("henryklimes/google-trends").call(run_input={
    "mode": "compare",
    "keywords": KEYWORDS,
    "geos": ["US"],
    "timeRange": "today 1-m",
    "includeRegions": False,
    "includeRelatedQueries": False,
})
for row in client.dataset(compare["defaultDatasetId"]).iterate_items():
    print(row["rank"], row["keyword"], row["average_interest"], row["trend_direction"])

# Related queries for each term (own 0-100 scale — not a ranking)
related = client.actor("henryklimes/google-trends").call(run_input={
    "mode": "keyword",
    "keywords": KEYWORDS,
    "geos": ["US"],
    "timeRange": "today 1-m",
    "includeRegions": False,
    "includeRelatedQueries": True,
})

A free Apify account includes monthly platform credit. The Actor charges $0.25 per 1,000 keyword-rows that return data (no_data and failed rows are free). These two runs billed 16 keyword-rows, so $0.004 in Actor events, plus about $0.001 of platform usage.

Limits of this snapshot

  • Google Trends is relative interest, not search volume.
  • One country (US) and one month. A worldwide view, or a 12-month window, would be a different picture.
  • The set is eight named queries, not "the AI tools market". Leaving out ChatGPT would make the smaller terms readable on a tighter scale; we kept it in so the gap is visible.
  • Related topics were not requested. Related queries include news events and typo-adjacent searches.
  • I built the Actor this report uses. The numbers are still just what the two runs returned.