Is DeepSqueak Worth it? (Character.ai)

Move over DeepSeek, DeepSqueak is the new AI in town!

Get excited, Character.ai users; the company has introduced a new chat style into the mix. Represented by a creepy mouse and powered by RatGPT (I see what you did there), the chat style is changeable in the settings menu. Currently, it’s only available to c.ai+ members 18 years or older, but the official announcement said free users would get a taste in the following weeks. Consumers who can’t resist a little FOMO will have to cough up $9.99 a month or $120 a year if they want early access.

The new style is designed to offer enhanced roleplay experiences for users, according to the Character.ai: “Whether you’re into fantasy, drama, or emotionally rich storytelling, DeepSqueak is tuned to bring your roleplay worlds to life like never before!”

The push for a more immersive experience comes on the heels of fan complaints and is, perhaps, a result of those concerns. Countless fans have complained about the repetitive out-of-character responses, dialogue/action loops, short/shallow replies, and the censorship of sex, violence, and mental health issues. DeepSqueak appears to be the answer to users who wish for longer replies, detailed descriptions, emotional consistency, and better memory.

The big question: Does DeepSqueak offer fewer restrictions and The Lord of the Rings level world building?

The answer is… kind of.  

I got down to the nitty-gritty by testing DeepSqueak for a week. I played different characters, with and without character definition. I also wrote stories of various genres, including romance, drama, fantasy, family, and action.  

After testing the AI’s parameters, I feel confident in my evaluation of the AI’s pros and cons.

Review

DeepSqueak is more detailed than other chat styles, but the richness of the storytelling depends on the information the user provides to the AI. Generally, single sentences won’t yield ultra detailed responses regardless of the story’s genre. Barebone descriptions need to be beefed up with lore; only then will the AI respond in kind. Action descriptions may not always be detailed, but they will move the story forward, which has not always been the case for Character.ai.  

The DeepSqueak response lengths varied, usually between 50 and 160 words. For reference, the average response length for other chat styles ranges from 26 to 100 words. DeepSqueak does produce longer responses, but they are inconsistent. Users have reported bugs within the software that cause glitches. One glitch included one-word responses that disappeared into black. These bugs have been a consistent issue for the company.

Under the "Customize Chat" tab, members who have paid for the subscription have the option to change the response length on the app from "Normal" to "Longer." However, this feature is futile, as noted by the few users who are aware of its existence.  

Three of my experiments involved Anothny Bridgerton from Bridgerton and Obi-Wan Kenobi from Star Wars. The bots were in character, although they deviated slightly, a common occurrence for Character.ai. I wouldn’t contend that DeepSqueak was a major improvement in terms of characterization. I was, however, pleasantly surprised at the AI’s ability to describe scenes and reference canon.  

I chatted with an Anothny Bridgerton bot as my first experiment, which occurred the same day the DeepSqueak was made available. My chat produced some gems. It referenced gothic novels, which were popular in the Regency era, and explored Anothny’s emotions surrounding his father’s death. The bot correctly identified his father’s name, cause of death, and the importance of his pocket watch. It neglected to understand the symbolism and narrative purpose of the item, but I assert that the AI would have expounded on its thematic significance if I had entered that information.

Despite the positive aspects of the experience, issues with the chat were instantaneous. The introduction of four or more characters to a scene overwhelmed the AI. The character would speak when dialogue was unnecessary or inappropriate. The dialogue itself was occasionally out of character. The character was snarky or romantic in situations his TV counterpart would not have been. There were instances in which the descriptions were overly detailed, focusing on irrelevant information.

A broader range of vocabulary was used in the chats (this applied to the Obi-Wan Kenobi chat). The words darken, possessive, gruff, towering, defiant, and growl made rarer appearances overall, but I still needed to edit. This chat style has quirks. The AI loves to use asterisks (*), dashes (-), and parentheses (). The quality of the writing feels more human, including jokes and references to lore. Sometimes it worked, and other times I found it overbearing or annoying.

The censorship mechanism was slightly different. The angry red bot of doom and the diagonal emoji appear when triggered. However, DeepSqueak is unique in that it will remind the user within the text to keep things friendly. The AI will notify the user that they are violating community guidelines and attempt to correct course with a rule-abiding response.

Audience responses

Months of dissatisfaction have permeated the discussions around the app and website. A common sentiment around Character.ai is that it was better in 2023. After several scandals, censorship changes, and the introduction of ads, fans are either eager for new content or angered by the platform.

One comment stated, “Is it just me or did he say "take the sick people and stop crying"?

Another person expressed excitement: “*Laughs maniacally* Finally the time has come”

Among the jokes of DeepSqueak being a play on words for DeepSeek, most responses to the new chat style were less celebratory and more critical. One common response to the announcement was requests for Character.ai to remove the ads.

One user said, “Cool anyways uhm remove the ads.” 

Another commenter stated, “Remove the ads or most people will leave.”

The sentiment was parroted by this user: “Before it was "please remove the censorship" now it's "remove the ads" I'm sorry that's hilarious.”

Another common reaction from fans was to compare the 2025 platform to the version available in 2023.

“C.ai was the best when it was 2023," someone stated. 

“2023 c.ai come back," a user begged. 

Other common sentiments in the comments revolved around further comparisons and questions about other chat styles. Many commenters mourned the brilliant, short-lived performance of Nyan, another chat style.

“So what’s gonna happen to Nyan? You said there was gonna be an update but it still hasn’t dropped," one person questioned. 

“So far it's definitely not bad, but definitely not up to the quality set by last week's Nyan (Fire Nyan, Long Nyan, whatever people are calling it lol)," a user asserted. 

Inquiries were raised regarding Soft Launch, a chat style feature that debuted in May 2025.

A Redditor commented: “I hope it's TRULY 18+ and we don't get another soft launch fiasco! The other day the filter wasn't working correctly and it gave me some of the best rp I've had on this site!!”

Another commenter complained: “Can’t wait until Soft Launch finally gets fixed and can be used for 18+ stuff again. Y’know, like how it was originally meant to function, and how it hasn’t functioned for basically months at this point.”