Is Character AI Losing Users and Money?

Yes on both counts. Character AI filed for Chapter 11 on October 12, 2026, while running an operating loss of about $8 million to $9 million a month. Its filing also puts monthly users at about 12 million, down from roughly 20 million in 2024.

In August 2024, Google’s licensing deal bought out Character AI’s investors at a reported $2.5 billion valuation. The opening bid in the bankruptcy sale is Disney’s $20 million, for a license to some of the technology.

The valuation drop is only half of the story. You can watch the same decline play out in the product itself, month by month.

Every new restriction, every paywall, every in-chat ad is a signal that something went wrong internally long before users noticed.

This article breaks down how Character AI got here and why 8 million users stopped showing up. It also covers what that means for you if you still use the platform or are wondering whether to keep paying for it.

If the limits have already pushed you to look elsewhere, Candy AI sells unlimited text chat on its paid plans.

How Character AI Went From $2.5 Billion to Chapter 11

Character AI spent far more than it earned every month, and a sale search that began in 2025 found no workable buyer, so it filed for Chapter 11 to sell itself under court supervision. In August 2024 it was riding high on a reported $2.5 billion valuation. By August 2025 it was in talks at a valuation of more than $1 billion, and reports said the company was exploring both a fundraise and a potential sale.

That’s a drop of up to 60% in about a year.

The causes aren’t mysterious. Running large language models at scale is brutally expensive, and Character AI’s revenue, up from about $7.4 million in fiscal 2024 to about $23.6 million in fiscal 2025, never came close to covering it.

Its first-day bankruptcy filing, laid out in Bondoro’s case summary, spells out the gap. Operating costs averaged about $18.5 million a month in 2025, roughly $10 million of it compute, down from about $13.8 million of compute a month in late 2024. Monthly revenue reached only $5.3 million by August 2026, even after ads arrived in mid-2025 and in-app purchases in early 2026.

Costs fell to about $13.3 million a month by August 2026, but net cash burn was still about $8.2 million a month, with lawsuits and regulators adding pressure. A banker hired in August 2025 approached about 28 potential buyers and got one proposal that was never actionable.

On October 12, 2026, Character AI filed for Chapter 11 to run a court-supervised sale, listing assets and liabilities each between $50 million and $100 million. Disney is the lead bidder at $20 million, for a non-exclusive license to some of its technology.

The app keeps running while the sale plays out, but it is not part of Disney’s bid. Character AI says the platform will be wound down if no offer to keep it running comes in. Our Character AI Chapter 11 guide tracks the sale dates and what happens to your chats.

The Numbers Behind the Decline

Character AI had about 20 million monthly active users in 2024, a figure Sacra and TechCrunch both reported. Its Chapter 11 filing puts the count at about 12 million, or 8 million fewer monthly users.

Not all of them left by choice. On November 25, 2025, Character AI ended open-ended chat for under-18 users, who were about 10% of its monthly users that October.

For context, Character AI still generates serious engagement among the users who stay: the company reports about 120 minutes a day per user on average, which beats most social apps.

Demand for AI companions held up. The product kept changing in ways that frustrated the exact users who were most invested in it.

What Put the Company Under Pressure

Three things compounded at once.

First, the AI companion space got a lot more competitive. Platforms improved meaningfully, giving users real alternatives that didn’t exist two years ago.

Second, compute stayed expensive. It averaged about $10 million a month in 2025, roughly five times the company’s monthly revenue that year.

Third, the team made a series of product bets that didn’t pay off, and those bets cost money that the company couldn’t easily recover.

The Feature Bloat That Drove Users Away

Here’s what happened to Character AI between 2023 and 2025: the team looked at their user base, a group of people who came for immersive one-on-one character conversations, and decided to build a social media platform.

Scenes, stickers, image generation, a discovery feed, collaborative features. Someone decided these would broaden the audience, though users had not asked for them.

Instead, the new features bloated the product, drove up server costs, and made the actual chatting experience feel like an afterthought.

What Users Kept Asking For Instead

The community feedback on this has been consistent for well over a year. Users wanted better memory across conversations and longer, more coherent responses.

They wanted the bot to stay in character reliably, and they wanted more control over persona settings.

None of those things required building image generation infrastructure. All of them would have made the core product meaningfully better.

In my experience watching platforms make this call, when you build for a hypothetical new audience while neglecting the people already using your product every day, you tend to lose both groups.

The Costs That Ate Through the Runway

Image generation and video are among the most computationally expensive features in AI. Character AI added both. Most users didn’t want them or didn’t use them regularly.

The infrastructure costs were real regardless. That’s a significant strategic misfire, and it shows up directly in the valuation numbers.

What makes it worse: those features could be removed without users complaining. The daily caps on swipes that arrived in March 2026, on a feature people relied on, created a genuine backlash.

The expensive features nobody wanted are still there.

What Character AI’s Monetization Means for Free Users Today

Free accounts still get unlimited messages, but they now also get ads, slow mode at busy times and, since March 2026, daily caps on swipes, go-ons and voice memos. Once the company started feeling the financial pressure, the product changed fast, starting with in-app ads in mid-2025.

A Charms currency followed, and in March 2026 Character AI added daily caps on swipes, go-ons and memos, which its announcement also calls playbacks. Charms no longer buy past those caps: the Charms FAQ, last edited June 17, 2026, lists only extra Imagine images or chats, skipping slow mode and bypassing ads.

These changes landed in quick succession, and the pile-up over a few months is what made the backlash so sharp.

To understand what’s changed for free users, here’s how the experience compares now:

What the Shift Feels Like in Practice

If you used Character AI two years ago and picked it back up today, the change is jarring. What was once a smooth experience now interrupts itself constantly.

Ads appear mid-session. Regenerate often enough and a pop-up tells you your daily swipes have run out. At busy times of day, slow mode kicks in and free-tier replies crawl.

Before (2023 free experience): Open a chat, write a message, get a response, hit regenerate if you want something different. Unlimited, smooth, no friction between you and the character.

After (2026 free tier): Open a chat, get a few exchanges in, see an ad. Swipe through enough replies and the daily cap stops you until it resets at 00:00 UTC. The pop-up pitches c.ai+ or the cheaper lite plan, though only c.ai+ raises the cap.

That shift in experience is why users are leaving: the product made itself harder to enjoy.

For a full breakdown of how the charms system works, this explainer on C.ai charms covers the token economy in detail.

Is C.ai+ Worth Paying For Now

The honest answer depends on how much you use the platform.

If you spend an hour or more on Character AI daily, the $9.99-a-month c.ai+ plan restores most of what the free tier took away. You get the highest swipe and go-on limits, no ads, no slow mode and the premium chat styles. The $4.99 lite plan trims ads and slow mode but keeps the free daily limits.

The bankruptcy adds one more factor. Character AI says subscriptions stay active and billing is unchanged for now, but a plan bought today could end early if no buyer keeps the platform running. A monthly plan carries less risk than the $94.99 annual one.

If you’re a casual user who drops in a few times a week, the free tier’s limits may not hit you much. The issue is that Character AI has now priced its best experience at the subscription level.

The free product is a deliberately degraded version designed to push you toward payment. That’s a legitimate business model, but it’s a different product than what people fell in love with in 2022 and 2023.

The full breakdown of C.ai+ versus the free plan lays out exactly what you gain and lose at each tier.

Where the 8 Million Missing Users Went

Plenty of the users who left kept using AI companions elsewhere. Some went to platforms with fewer restrictions and a better free experience.

Others found tools that fit their specific use case more precisely.

The five reasons users most often cite for leaving Character AI:

1. Daily swipe and go-on caps cut off long-running roleplay or story sessions mid-flow
2. Slow mode at busy times, with Charms or c.ai+ sold as the way to skip the wait
3. In-chat ads broke immersion at the worst possible moments in a conversation
4. Age checks from late November 2025, which asked adults the system flagged as possible minors for a Persona selfie or ID
5. The platform kept removing features rather than fixing the ones that were already broken

That fifth point is the one that stings most in the community. Chat styles are the clearest case: Character AI retired Roar, Pawly, Goro and Dynamic on April 28, 2026, and its Legacy menu says the older styles left there will be retired over time.

When a platform removes useful features without replacing them with something better, trust erodes quickly.

For users who want the core experience of character-driven AI chat without the restrictions, there are solid alternatives now.

Candy AI keeps persistent characters across sessions, though its free tier stops at 5 messages in total. Nectar AI is worth looking at if you want deeper character customization.

The broader trend in this space, and where the community has been heading, is tracked in our Character AI user exodus piece.

Should You Still Use Character AI in 2026?

You can still use Character AI today, because the app keeps running while its Chapter 11 sale plays out, but nobody can promise it survives the sale. It still has about 12 million monthly active users, more than 10 million user-created characters, and one of the deepest libraries of AI character personas available anywhere.

Character AI expects the sale to finish by the end of November or early December, so the platform’s fate should be clearer by then. Exporting your chats now costs nothing and protects you whichever way the sale goes.

The character library is a genuine asset, yet the platform kept making its own offer harder to accept.

The valuation drop, the user decline, and the ongoing feature removals all pointed to the same squeeze: converting a massive free-user base into paying subscribers while compute costs ran far ahead of revenue.

Whether the app survives now depends on the sale, and on whether a buyer thinks those users are worth the compute bill.

Here’s a plain breakdown of where each type of user stands right now:

My read on it: Character AI built something people genuinely loved, then spent the next year making decisions that worked against the thing that made it worth loving.

Character AI’s own choices, made while compute cost several times its revenue, pushed users toward competitors with fewer restrictions.

Whether the platform recovers now depends on who buys it, and on whether they can find a cost structure that works without making the product worse every quarter.

That’s not certain. And for users who are tired of waiting, the alternatives have gotten good enough that waiting isn’t the only option.