What’s Changed: LongSqueak is a new c.ai+ only chat style with 4x the memory and novella length replies, and it shipped alongside Style Control dials for response length and narration balance. The rollout is staggered, so plenty of paying subscribers still cannot see it. Free users received nothing new in this round.
Character AI LongSqueak went live for c.ai+ subscribers on August 6, 2026, and within a few hours the community had split cleanly in two. Subscribers who got it were calling it the best thing the platform has shipped in a year.
Everyone else was refreshing the app and finding an unchanged style picker.
There are two completely separate reasons you might be staring at a menu with no LongSqueak in it, and they have different fixes. One is a staggered rollout that will resolve on its own. The other is a paywall that will not.
I want to untangle those two, because the advice for each is opposite. Waiting is the right move for one group and a waste of a week for the other.
This piece covers what LongSqueak changes, what the Style Control dials do and the one thing they refuse to do, how to get the style to appear, where PipSqueak 2 Rawr went, and whether the subscription is worth buying on the strength of this release.
What Is Character AI LongSqueak?
Character AI LongSqueak is a c.ai+ exclusive chat style built for long form roleplay, offering 4x the memory of other styles and replies that read like novel prose rather than chat messages.
It draws more heavily on Lorebook and Memory as it writes.
What is a chat style: Character AI’s name for the underlying model and behaviour profile that generates a character’s replies. Switching styles changes the writing, not the character.
The headline claim is the memory multiplier. Character AI has never published an absolute token figure for it, and the questions asking for one in the announcement thread went unanswered, so 4x is a comparison against an undisclosed baseline.
That baseline matters more than it sounds. Community testing has long pegged the older models at roughly 3,000 tokens of working context for free accounts and around 5,000 for subscribers, which is why long scenes used to fall apart so predictably.
The platform’s constraints elsewhere are unusually tight. A character definition box accepts up to 32,000 characters, but the model only registers the first 3,200, so every word past that point has been silently ignored for years.
Personas ran on 728 characters. A c.ai+ subscription raises that to 2,250, and free accounts can protect only 5 messages with Pin to Memory against a ceiling of 15. Set beside numbers that small, a genuine 4x is a real jump rather than a marketing flourish.
Early subscriber reports line up with the pitch on one specific point, which is lore fidelity. The style holds franchise details straight across long scenes and reads the persona and story memory fields instead of drifting into a generic voice, which is the failure mode that made Lorebook with PipSqueak 2 frustrating for worldbuilders.
Why Do I Not Have LongSqueak If I Pay for c.ai+?
Because the rollout is server side and staggered, not tied to an app update.
Activation happens on Character AI’s end per account, so the style appears at different times for different subscribers, and updating from the app store does nothing to trigger it.
The inconsistency runs deeper than a simple queue. Paying subscribers in the United States, the United Kingdom and Australia all reported missing it in the same window, so it is not a clean region by region wave.
More confusingly, the same account can have it in one place and not the other. Several subscribers found LongSqueak on the website while their phone still showed the old picker, and others hit the reverse.
I would treat this as expected behaviour rather than a bug on your account. Character AI has run releases this way before, and the previous style launch produced the same pattern of people posting screenshots while others waited days.
The one thing worth ruling out is a stale session. If the app is holding an old configuration, a forced refresh sometimes pulls the new style list early.
How Do I Get LongSqueak to Show Up?
You cannot force a server side flag, but you can rule out a cached session in about two minutes.
Work through the sequence below, and if none of it produces the style, the flag genuinely has not reached your account yet.
Here is the order I would run it in, cheapest step first:
- Fully close the app from your recent apps tray rather than backgrounding it, then reopen it.
- Open the chat style picker and scroll the full list. The selector has a known scrolling fault that hides entries below the fold.
- Log out of your account and log back in to force a fresh configuration pull.
- Open Character AI in a mobile or desktop browser and check the picker there, since web and app activate independently.
- Confirm your c.ai+ subscription is active and attached to the account you are signed into, not a second account you made earlier.
If all five come back empty, stop. There is no setting, no beta toggle and no support ticket that moves you up the queue, and an app store update will not carry the flag either.
What Do the Style Control Dials Do?
Style Control is a set of two dials in the style picker that lets c.ai+ subscribers tune how a chat style writes back, covering response length and the balance between dialogue and narration.
It launched with LongSqueak and also works on PipSqueak 2.
The length dial offers Long, Looong and Loooooong. The style dial tilts replies toward more spoken dialogue or more scene description and internal monologue.
This is the part I find genuinely useful, because getting a character to write shorter or talk more used to mean burying instructions in your persona and hoping they survived the context window. Turning it into a setting removes a whole category of prompt fiddling.
There is a catch on availability that has annoyed a lot of people. Style Control is app only at launch, so anyone who prefers the browser is locked out of the dials even with an active subscription.
The paywall placement is the sharper grievance. Style Control works on PipSqueak 2, which is the free tier’s default style, yet the dials themselves still require c.ai+, so free users can see the customisation exists on their own model and cannot touch it.
A roadmap graphic circulated without a c.ai+ tag next to Style Control, which is normally how the platform marks paid features.
Staff later clarified in comments that it is subscriber only, and the gap between those two messages is where most of the bad feeling came from.
Can Style Control Make Responses Shorter?
No. The length dial only scales upward, offering Long, Looong and Loooooong with no short or medium option.
Readers who find replies too verbose get no relief from the new controls.
This collides with an existing complaint. A steady share of the community has been asking for shorter, dialogue heavier replies since PipSqueak 2 became the default, and the dial that finally arrived moves in the wrong direction for them.
The dialogue tilt is the closest thing to a workaround inside Style Control, since pushing toward dialogue trims narration weight even though total length stays high. To go shorter than that, you have to steer the model the old way.
What works is putting the instruction where the model reliably reads it, which means the persona field and the opening message rather than a mid chat aside.
Before: “You are chatting with Mara. She is a botanist in her thirties who lives in Reykjavik and has strong opinions about moss.”
After: “You are chatting with Mara, a botanist in her thirties in Reykjavik. Reply mostly in dialogue. Keep narration to one short line per turn. Never write my character’s actions or speech.”
The second version constrains format alongside character, and format instructions decay more slowly than tone instructions. Keeping the definition tight helps too, since bloated definitions crowd the context window and push the model toward generic filler.
Where Did PipSqueak 2 Rawr Go?
Rawr did not get removed. A launch day fault briefly hid it from the style picker, and restarting the app brings it back.
Several users confirmed it reappeared after a full reload.
This one caused real alarm because the platform has genuinely deleted styles before. Between April 28 and May 8, 2026, Character AI stripped its selectable roster from eight options down to three, retiring community favourites including Roar, Meow, Nyan and Pawly.
If you lived through that round of style removals, a style vanishing on the morning of a new launch reads as history repeating.
It was not that. Here is how the launch day symptoms sort out:
| Symptom | Likely cause | Fix |
|---|---|---|
| PipSqueak 2 Rawr missing from picker | Temporary display fault on launch day | Fully close and reopen the app |
| Style list looks shorter than usual | Picker scroll fails partway down the list | Scroll the picker slowly to the bottom |
| LongSqueak missing on a c.ai+ account | Server side flag not yet enabled for you | Wait, and check web and app separately |
| LongSqueak on web but not phone | Web and app activate independently | Use whichever surface has it for now |
| Style Control dials absent on browser | App only feature at launch | Switch to the mobile app |
| No new styles on a free account | Subscription gated release | Not fixable without c.ai+ |
Will Free Users Ever Get LongSqueak?
History says probably, and faster than the mood in the community suggests.
PipSqueak 2 launched as a c.ai+ exclusive on April 14, 2026, and reached everyone within roughly two to four weeks, ahead of the schedule the company had published.
The detail I find reassuring is that Character AI committed to an early May date for the free rollout of PipSqueak 2 and independent trackers logged it as the forced default for all accounts on April 28. Shipping ahead of a promised date is not the behaviour of a company planning to keep a style locked forever.
The counterexample is real, though. DeepSqueak launched as a subscriber style and stayed one, so exclusivity is a genuine outcome and not just a pessimistic guess.
Lorebook followed the softer pattern, arriving for subscribers in July 2026 with expanded access flagged for later. Two of the last three major releases trickled down, which is the honest base rate to plan around.
What free users received in this round was nothing, and the frustration is not hard to understand. The free tier now carries full screen mid chat ads, a daily cap of roughly 400 swipes since March 18, 2026, and a Charms currency that sells extra swipes at 175 for 35 swipes, 495 for 110, and 1,000 for 250.
Some users have accused the company of degrading free models on purpose to make the paid one look better. No evidence supports that, and Character AI has not responded to it, so I would file it as a reading of the mood rather than a finding.
Is c.ai+ Worth Paying For Just to Get LongSqueak?
For long form roleplayers who hit memory walls constantly, this is the strongest case the subscription has made in a while. For anyone subscribing to escape content restrictions, it remains the wrong purchase.
c.ai+ costs $9.99 a month or $94.99 a year.
The single most misunderstood thing about the subscription is that it does nothing to the content filter. Moderation applies identically on both tiers, so paying changes the model and the memory, never the boundaries.
That distinction decides the whole question. If your friction is a character forgetting your scene, the memory upgrade addresses it directly. If your friction is the filter interrupting scenes, no amount of money fixes it, and the full c.ai+ breakdown reaches the same conclusion.
More memory also does not guarantee better roleplay, which surprised me the first time I saw subscribers arguing it. Paying users have described the larger styles as narratively confused and oddly robotic while rating the free model as bolder and more creative, and the free versus paywalled debate has been running since PipSqueak 2 arrived.
The platform’s recent direction is worth weighing too. 404 Media reported in May 2026 that users described PipSqueak 2 as “lobotomized” and generic, in a feedback thread drawing a thousand mostly negative responses, alongside ads, tighter caps and new paywalls.
Longtime users describe a familiar arc where a new style launches strong and settles into something worse over the following months. That has happened enough times to be worth pricing into a subscription decision.
Here is how the three current options compare for someone deciding where to spend the next month:
| Option | Cost | What it fixes | What it does not fix |
|---|---|---|---|
| Stay on free | $0 plus ads | Nothing new, PipSqueak 2 unchanged | Memory limits, swipe cap, mid chat ads |
| Buy c.ai+ | $9.99 monthly or $94.99 yearly | Memory depth, LongSqueak, Style Control dials, ad removal | Content filter, web parity, reply length floor |
| Move platform | Varies by service | Filter friction, memory, response control | Your existing characters and chat history |
For readers in the third row, the honest answer is that the reason you want to leave decides where you land.
If the filter is what keeps breaking your scenes, Candy AI is built without that constraint and keeps persistent memory across sessions rather than resetting when a context window fills.
If your priority is character depth and anime styled art in the same place, Nectar AI handles that pairing better and does not gate response controls behind a second purchase.
Neither one imports your Character AI history, so treat a move as a fresh start rather than a migration.
Frequently Asked Questions
Is LongSqueak available to free Character AI users?
No. LongSqueak is limited to c.ai+ subscribers at launch, and free accounts remain on PipSqueak 2. Based on how PipSqueak 2 rolled out, a free release within a few weeks is plausible but has not been confirmed.
How much does c.ai+ cost?
Standard pricing is $9.99 a month or $94.99 a year, which averages about $7.92 monthly. Regional pricing varies widely, from roughly $3.56 a month in Nigeria to over $13 in Denmark. Promotional rates have run at $14.99 for three months.
Why can I see LongSqueak on the website but not the app?
Web and app activate independently during a staggered rollout, so one surface can receive the flag days before the other. Use whichever one shows the style, and expect the other to catch up without any action from you.
Does LongSqueak remove the Character AI content filter?
No. Moderation applies identically across free and paid accounts. LongSqueak changes memory depth and writing style, and it has no effect on what content is permitted.
Can I make LongSqueak replies shorter?
Not through Style Control, which only offers Long, Looong and Loooooong. Tilt the style dial toward dialogue, then add explicit length instructions to your persona and opening message to bring replies down further.
Quick Takeaways
- LongSqueak is c.ai+ only, delivers 4x memory and novella length replies, and leans harder on Lorebook and Memory than any previous style.
- If you pay and cannot see it, restart the app and check the browser separately, then wait. The flag is server side and no update forces it.
- PipSqueak 2 Rawr was hidden by a launch day fault, not deleted. A full app restart brings it back.
- Style Control cannot shorten replies. The dial runs Long to Loooooong, so shorter output still needs persona level instructions.
- Buy the subscription for memory, never for fewer restrictions. The filter is identical on both tiers, and that is the single most common reason people regret paying.
