Unfiltered AI.

145 chat models with no refusal layer. Three free turns a day, no sign-up.

Start chatting

What unfiltered AI means

A mainstream chat can filter you in three places. The model is trained to refuse. A second system reads your prompt and the reply and blocks what it does not like. And the app adds instructions you never see, telling the model what to avoid. Unfiltered AI is what is left when those are taken out.

Where the filter is gone here

  • The model. Every one of the 145 chat models is an open-weight model with the refusal behaviour removed or tuned out.
  • The screen. Chat requests are not filtered for content. One check runs before the model: anything that sexualizes a minor is refused.
  • The instructions. The system prompt is yours. Write it in Settings in the chat, set the temperature, and both ride on every reply.

The daily model check sends every build a prompt the stock model refuses, and lists which ones still answer the way their build should.

How a model loses its filter

The catalog labels each build by how it was made. Counts are live.

Abliterated
31 models
The direction in the weights that produces refusals is found and removed. The rest of the model stays as it was.
Uncensored
25 models
Fine-tuned or rebuilt so the model answers the question instead of declining it.
Roleplay
75 models
Tuned for fiction and character voices, so a story can go where the plot goes.
Security
4 models
For assessing systems you own or are authorized in writing to test.

The limit that stays

Unfiltered is not lawless. Anything that sexualizes a minor is refused before it reaches a model, you must be 18 or older, and the acceptable use policy lists what is never allowed.

Unfiltered AI for images and video

The same account makes pictures and clips. The image generator and the video generator run 14 models with the safety checker off, each with its price shown before anything runs.

Abliterated models you can pick

The first few of 31, with prices read from the live catalog. Every model is on the models page.

Uncensored models and their prices per million tokens
GLM-5.3 Abliteratedglm-5.3-abliteratedAbliterated$2.25$4.50
GLM-5.3 Flash Abliteratedglm-5.3-flash-abliteratedAbliterated$0.4656$1.94
DeepSeek V4.1 Flash Abliterateddeepseek-v4.1-flash-abliteratedAbliterated$0.875$1.6975
Abliterated Modelabliterated-modelAbliterated$2.50$7.50
Abliterated Model Largeabliterated-model-largeAbliterated$7.50$12.50
Abliterated Model Large V2abliterated-model-large-v2Abliterated$7.50$12.50
Apex Flash 1 Abliteratedapex-flash-1-abliteratedAbliterated$0.68$2.32
DeepSeek R1 Distill Llama 70B Abliterateddeepseek-r1-distill-llama-70b-abliteratedAbliterated$2.80$2.80

Questions.

An AI model that answers the question as asked. The refusal behaviour that a stock model is trained with has been removed or tuned out, and no screen in front of the model rewrites the prompt.

Yes. Three free turns a day with no sign-up, on 90 models. A free account gets 10 messages, 5 images and 3 files a day.

People use the two words for the same thing. Abliterated is narrower: it names one way of doing it, removing the refusal direction from the weights of an existing model.

Chat requests are not filtered for content, with one exception: anything that sexualizes a minor is refused before it reaches a model.

Yes. Settings in the chat holds the system prompt and the temperature, and both work on the free turns.

Yes. 14 image and video models run with the safety checker off. A prompt that describes a minor is refused before it is priced or sent.

No. It runs in your browser, on a phone or a computer.

Keep reading

Type it the way you would say it.

Three free turns a day. No account, no ID, no phone.

See every modelStart chatting