Where are Character.AI users going? Search and app store data
Since Character.AI’s owner filed for bankruptcy on October 12, 2026, the public data does not show people leaving: weekly US searches for Character.AI are about the same as the 7 weeks before the filing. Of the alternatives, Zeta is growing much faster than the rest: its searches are 61% above the same weeks, and in the US App Store it gains about 15 times as many new ratings a day as Character.AI.
- US searches for Character.AI, week to Oct 3124,371−1% vs. the 7 weeks before the filing
- US searches for Zeta, same week8,467+61% vs. the 7 weeks before the filing
- New US App Store ratings a day, Oct 31–Nov 4Zeta 712Character.AI 48
Searches for Character.AI have not moved
Bing Webmaster Tools reports how many times a week people in the United States search for any phrase. We counted searches that contain each phrase, for Character.AI and for the apps people name as alternatives. In the 7 weeks before the filing, searches for “character ai” averaged 24,504 a week; in the week to Oct 31 there were 24,371.
- Character.AI
- Zeta
- PolyBuzz, Janitor AI, Talkie
The news itself faded fast: searches for “character ai bankruptcy” went from 318 in the week of the filing to 34 in the week to Oct 31. Searches for “character ai alternatives” are 21% above their earlier level, at 107 a week.
| Search phrase | Average week, Aug 29–Oct 10 | Week to Oct 17 | Week to Oct 24 | Week to Oct 31 | Latest vs. before |
|---|---|---|---|---|---|
| Character.AI “character ai” | 24,504 | 26,112 | 23,006 | 24,371 | −1% |
| Zeta “zeta ai” | 5,261 | 8,392 | 7,990 | 8,467 | +61% |
| Janitor AI “janitor ai” | 28,134 | 27,632 | 28,141 | 27,181 | −3% |
| PolyBuzz “polybuzz” | 101,527 | 97,857 | 88,258 | 82,439 | −19% |
| Talkie “talkie ai” | 9,345 | 9,684 | 8,846 | 9,281 | −1% |
| Emochi “emochi” | 6,477 | 8,414 | 7,031 | 7,366 | +14% |
| Kindroid “kindroid” | 1,900 | 1,895 | 2,045 | 2,076 | +9% |
| Perchance AI chat “perchance ai chat” | 2,690 | 2,678 | 2,570 | 2,412 | −10% |
| Chai “chai ai” | 1,437 | 1,367 | 1,300 | 1,157 | −19% |
| Replika “replika” | 485 | 551 | 420 | 407 | −16% |
| Nomi “nomi ai” | 388 | 517 | 475 | 375 | −3% |
| SillyTavern “sillytavern” | 317 | 395 | 353 | 347 | +10% |
| “character ai alternatives” | 88 | 83 | 94 | 107 | +21% |
| “character ai bankruptcy” | — | 318 | 63 | 34 | new |
| “is character ai shutting down” | — | 50 | 24 | 24 | new |
Zeta is growing faster than the other alternatives
Every app store shows how many people have rated an app. The count grows when someone new rates it, so new ratings a day are a public trace of people arriving. We read the counts once a day since October 31, 2026: in the App Store in nine countries and in Google Play in the United States.
In that time Zeta gained about 712 new US App Store ratings a day, 15 times as many as Character.AI (48 a day, with 561,037 ratings in total). In Google Play in the United States, Zeta added 1,221 a day and Character.AI 258. Zeta, an AI character app from South Korea that is big in Japan and Korea, was growing in the US before the case began: in the 7 weeks before the filing its searches were already 72% above the 7 weeks before that, and they stepped up again in the week of the filing. We cannot say how much of its growth comes from Character.AI users.
| App | App Store, US | Google Play, US | App Store, 9 countries | Average rating, App Store US |
|---|---|---|---|---|
| Zeta | 712 | 1,221 | 2,276 | 4.7 |
| Talkie | 98 | 416 | 120 | 4.7 |
| Janitor AI | 53 | — | 67 | 4.5 |
| Character.AI | 48 | 258 | 82 | 4.3 |
| Kindroid | 38 | 43 | 49 | 4.8 |
| PolyBuzz | — | 714 | — | — |
| Chai | — | 374 | — | — |
| Nomi | 2 | 2 | 2 | 4.6 |
| Replika | 0 | 11 | 0 | 4.4 |
What would change it
The app still works, nobody has announced a closing date, and people rarely move a long history before they have to. The dates to watch: bids are due on November 20, an auction follows on November 23 if a rival bid qualifies, and the sale hearing is on November 24. The court can move them; our status page keeps them current, with sources.
A closing date is what moves people. When Google said in 2013 that Reader would close, Feedly reported 500,000 new users in about two days. If the sale ends with no one running the app, we expect these numbers to move quickly, and we will update them.
How we measured
- Searches. Bing Webmaster Tools keyword statistics for the United States in English: weekly counts of searches that contain the phrase (Bing’s broad match), weeks ending on Saturday, read on November 5, 2026. “Before” is the average of the 7 weeks to Oct 10. Bing gives no row for a week it does not count; those cells are empty.
- Ratings. Apple’s public app lookup (the rating count of each app) in nine App Store countries, and the rating count on each app’s Google Play page in the United States, read once a day. New ratings a day are the change in the count divided by the exact time between the first and the latest reading.
- Apps. The apps that appear in lists of Character.AI alternatives and that people search for by name. The search phrase for each app is shown in the table.
What this data cannot tell you
- Bing is a small share of search in the United States, and searches count queries, not people.
- A rating is not an install. Apps ask for ratings in different ways and at different moments, so an app’s count is best compared with itself over time, and gaps between apps are a rough guide.
- Our app store readings start on October 31, 2026, after the filing, so the change across the filing itself is visible only in the search data.
- People who move without searching, or who stop using such apps at all, do not show up here.
Who made this
We make Amavie, an app of AI characters. It is one of the places people might go, so we have a stake in this question and we say so. Amavie is not in the tables or the files. Character.AI, Zeta and the other names are trademarks of their owners; we are not affiliated with them.
Use this data
- Weekly US searches, CSV and daily app store ratings, CSV.
- Free to use under Creative Commons Attribution 4.0: credit Amavie and link to this page.
- Cite as: Amavie, “Where are Character.AI users going? Search and app store data,” updated November 5, 2026, amavie.ai/where-character-ai-users-go.
- Questions about the data: support@amavie.ai.
- We read the stores every day and refresh this page after the bid deadline and after the sale hearing.
Frequently asked questions
Are people leaving Character.AI?
Not in large numbers so far. Weekly US searches for Character.AI in Bing are about the same as the seven weeks before the October 12, 2026 bankruptcy filing, and the app still gains new ratings every day.
Which Character.AI alternative is growing?
Zeta is growing faster than the other alternatives. Its US searches are 61% above the seven weeks before the filing, and in the US App Store it gained about 712 new ratings a day, 15 times as many as Character.AI (48). Its growth started before the filing, so not all of it comes from Character.AI users.
Has Character.AI shut down?
No. The app keeps running during the court-supervised sale. Bids are due on November 20, 2026, an auction follows on November 23 if a rival bid qualifies, and the sale hearing is on November 24. The company says the platform will be wound down if no bid keeps it running.
Can I use this data?
Yes. The tables and both CSV files are free to use under Creative Commons Attribution 4.0: credit Amavie and link to this page. The numbers come from Bing Webmaster Tools and the public App Store and Google Play listings.