What's New
Immersion is what separates ordinary conversations from memorable roleplay sessions that stay with you long after they end. ZeroFata’s Painted Fantasy Visage v4 is a model that proactively drives the story, adds depth to characters, and delivers an immersive AI roleplay experience. We tested the model with five different character…
People often write off older models since LLMs are evolving faster than most can keep up with. Sao10K’s L3 8B Stheno v3.2 proves that a well-trained older model can still outperform newer ones, especially when it comes to AI roleplay. We tested the model with five different character cards and…
People often criticize smaller models for making characters feel one-dimensional or for forgetting context and ruining the flow of a roleplay. They’re not perfect, but for those running LLMs locally with limited hardware, small models are their only option. IceFog72’s IceMoonshineRP 7B is one such model. It’s small, yet capable…
Don’t you just hate it when a stubborn character suddenly gives in or a ruthless one backs down without a fight? Nothing in AI roleplay is more frustrating than an LLM breaking character. TheDrummer’s Snowpiercer 15B v3 is a model that doesn’t yield. It refuses to break character just to…
Finding a model for AI roleplay that excels at portraying a character and being a proactive and creative narrative partner is a rare treat. TheDrummer’s Cydonia 24B v4.1 is one such model. This 24B fine-tune consistently demonstrated why it’s a fantastic narrative partner. We tested the model with five different…
We didn’t know AI could fall sick, and neither did Vedal, the developer of the AI VTuber Neuro-sama. She and her twin sister, Evil Neuro, took a couple of days off from streaming to recover from their illness. The Swarm, Neuro and Evil’s least dedicated fanbase, came together to wish…
WyvernChat Studio Editor is a visual editor for creators, where they can create WyvernScripts and commands and test them in real time. The Studio Editor empowers creators to make their characters more immersive and engaging, and it just got better. The developers have also improved WyvernChat’s UI, making it easier…
Running LLMs locally requires a desktop or laptop with decent hardware. Those who have a gaming or productivity rig with a dedicated GPU can run small to medium models locally without breaking a sweat. If you don’t have a dedicated GPU or have one with 6GB or less VRAM, running…
Running LLMs locally requires a desktop or laptop with decent hardware. Those who have a gaming or productivity rig with a dedicated GPU can run small to medium models locally without breaking a sweat. If you don’t have a dedicated GPU or have one with 6GB or less VRAM, running…
On September 26, 2025, Character AI issued a vague statement saying it is removing characters from its platform following a copyright notice. The statement didn’t specify which characters were on the chopping block. In the hours following the statement, it became clear that Character AI was removing Disney characters from…
DeepSeek has launched its latest model, DeepSeek V3.2. The experimental model builds upon V3.1 Terminus and introduces DeepSeek Sparse Attention (DSA). With it comes a 50% drop in the official API pricing. The open-source model is available for download on HuggingFace, and DeepSeek has updated its official API to use…
Chub is an online AI roleplay platform that gives users maximum control over their experience, including the ability to fork and export user-generated content, such as characters, lorebooks, and more. Other platforms often hide character definitions or lack options to fork or easily export user-generated content. They do this to…
Online AI roleplay platforms pop up on an almost daily basis, and they all follow the same formula. They mimic other established platforms, offer limited features, restrict free usage, benefit from user-generated content, create dependency, and push recurring subscriptions to cash in on the trend. There are a few exceptions,…
WyvernChat offers its users four free LLMs with no usage limits or filters. The free models are fine-tuned for AI roleplay and perform decently in most scenarios. However, their context size is limited, and you might eventually find their responses becoming repetitive or predictable. If you’re bored of using WyvernChat’s…
Chub stands at the forefront of uncensored online AI roleplay. It’s a sanctuary for many who have moved from more restrictive platforms. The platform’s policies, combined with its approach of giving users control over all user-generated content, have made it the de facto bridge between local and online AI roleplay.…
JanitorAI is one of the most popular online AI roleplay platforms, with over 2 million daily users. The platform’s unfiltered and free LLM, ease of use, and its community of creators contribute to its success. Despite its success, JanitorAI does not have an app. The developers chose not to prioritize…
During AI roleplay, every message you send to the LLM is one big prompt that includes the character definition, scenario, system or custom prompts, conversation history, and more. As your conversation with the LLM progresses, the amount of repetitive input also increases. What Are Input Tokens? When you roleplay with…
SillyTavern is a free, open-source frontend for interacting with LLMs that you can run locally on your device. It’s a powerful, versatile tool that you can customize to match your needs, as long as you’re willing to embrace its learning curve. The developers don’t position it as a simple solution…
SillyTavern is a free, open-source, versatile frontend designed for power users. The developers offer a powerful sandbox tool that you can customize as you like, and they are unapologetic about its steep learning curve. But not everyone who uses SillyTavern for AI roleplay needs to know how to develop extensions,…
Many people struggle to set up and use SillyTavern despite its comprehensive official documentation and various community-created guides there to assist them. SillyTavern is a versatile frontend designed for power users. Its developers do not present it as a simple product tailored for any specific use case and instead encourage…
A model that’s accessible, context-aware, and decent at both chat and adventure AI roleplay sounds too good to be true. But Sicarius’ Impish LLAMA 4B promises to deliver exactly that. Despite its small size, it shows surprising awareness and impressive performance in AI roleplay. We tested the small yet capable…
Hot on the heels of the platform announcing that developers are prioritizing UI and Quality of Life updates, JanitorAI introduces Scripts and Advanced Generation Settings. The platform now also has a logo that consistently represents its brand across all forms of media, from web to print, and is flexible enough…