Models / Uncensored AI models
Uncensored AI models
Models tuned to follow your instructions with few or no refusals: Dolphin, Hermes and abliterated versions of Llama, Qwen and Gemma. They run privately on your iPhone or iPad in QUIET Ai. QUIET Ai is rated 18+; use them responsibly.
The smallest Dolphin: under 1 GB, so it runs on nearly every iPhone that QUIET Ai supports.
Dolphin's few-refusals tuning on Qwen2.5 3B. Quick on any recent iPhone and good in many languages.
Qwen2.5 3B with refusals removed. Small, fast and multilingual.
The small Dolphin. Same few-refusals tuning, light enough for most recent iPhones.
The compact Hermes 3 for stories and characters on a phone with less memory.
Microsoft's Phi-4 mini with refusals removed. Strong at reasoning and maths for its size.
Qwen3 4B with refusals removed. Step-by-step thinking in a size that suits 8 GB iPhones.
Google's Gemma 3 4B with refusals removed by Maxime Labonne. Small, fast and multilingual.
A Mistral 7B tuned by Nous Research for open conversation and structured answers. Apache 2.0, so there are few strings attached.
The classic Dolphin on Mistral 7B: follows the system prompt, with a 32K context for long chats.
Mistral 7B v0.3 with refusals removed. A dependable all-rounder under Apache 2.0.
Qwen2.5 7B with refusals removed. A strong all-rounder in more than 25 languages.
The Qwen2.5 coding model without refusals: security research, scripts and code the original declines to write.
A general assistant tuned to follow your system prompt instead of a built-in rulebook. Few refusals, strong at writing and code.
Dolphin 3.0 with its remaining refusal behaviour removed by abliteration, a technique that edits the model's weights rather than its prompt.
Nous Research's steerable assistant: long conversations, characters and creative writing, and it sticks to the system prompt.
Meta's Llama 3.1 8B with refusals removed by Maxime Labonne. One of the most downloaded abliterated models.
The DeepSeek R1 reasoning model (Llama 8B distill) with refusals removed. Shows its thinking before it answers.
Qwen3 8B with refusals removed by abliteration. Keeps Qwen3's step-by-step thinking and many languages.
Qwen3 8B abliterated and then tuned to be a direct, helpful assistant. Thinks step by step.
Questions
What is an uncensored AI model?
An open model that has been tuned, or edited, to follow your instructions instead of refusing. Dolphin and Hermes are trained that way; abliterated models have the refusal behaviour removed from an existing model such as Llama, Qwen or Gemma.
Can I run an uncensored AI on my iPhone?
Yes. QUIET Ai runs these models on the iPhone or iPad itself. Models of 3B and smaller run on most recent iPhones; 7B and 8B models want an 8 GB or 12 GB iPhone (iPhone 15 Pro and later). QUIET Ai checks your device before downloading.
Is it private?
The model runs on your device and works in Airplane Mode. Your chats are not sent to a server, there is no account, and nobody reviews what you type.
How do I add one to QUIET Ai?
Open Models, then More, then Add from Hugging Face. Paste the repository name shown on the model's page and pick the Q4_K_M file.
Is it allowed?
These are openly published models with their own licences, linked on each page. QUIET Ai is rated 18+. You are responsible for what you make with them and for following the law where you live.