Uncensored LLMs
227 uncensored language models, most downloaded first: the model each is based on, how its limits were removed, and where to get it.
AIModelFind lists these models from public sources; it does not host or run them. Check each model's license and the law where you use it.
| Model | Based on | Method | Download as | Downloads / mo | Hosted API |
|---|---|---|---|---|---|
N Gemma-3-27B-it-NP-Abliterated by Nabbers1999 | Gemma 3 27B | Fine-tuned | GGUF | 1K | Weights only |
 Aion-RP 1.0 (8B) by Aion Labs | Llama 3.1 8B Instruct | Fine-tuned | — | — | $1.60/1M 2 providers |
Uncensored by Cognitivecomputations | — | Fine-tuned | — | — | $0.90/1M 2 providers |
 Abliterated Model Large Flash by Abliteration.ai | — | Abliterated | — | — | $2.50/1M Venice AI |
Abliterated Model Large V2 by Abliteration.ai | — | Abliterated | — | — | $5/1M Venice AI |
 Gemma 4 Uncensored by Venice | — | Uncensored | — | — | $0.50/1M Venice AI |
 GLM-5.3 Uncensored by nanogpt | GLM-5.3 | Uncensored | — | — | $2.25/1M NanoGPT |
Olafangensan GLM-4.7 Flash Heretic by Venice | GLM-4.7-Flash | Abliterated | — | — | $0.40/1M Venice AI |
Qwen3.8 27B Uncensored by nanogpt | Qwen3.8 27B | Uncensored | — | — | $3/1M NanoGPT |
Venice Uncensored 1.2 by Venice | — | Fine-tuned | — | — | $0.90/1M Venice AI |
Venice Uncensored Role Play by Venice | — | Fine-tuned | — | — | $2/1M Venice AI |
Abliterated Model by Abliteration.ai | — | Abliterated | — | — | Weights only |
Abliterated Model Large by Abliteration.ai | — | Abliterated | — | — | Weights only |
 DeepSeek R1 Llama 70B Abliterated by huihui-ai | — | Abliterated | — | — | Weights only |
DeepSeek R1 Qwen Abliterated by huihui-ai | — | Abliterated | — | — | Weights only |
Dolphin 2.6 Mixtral 8x7b by Cognitivecomputations | Mixtral 8x7b Instruct | Fine-tuned | — | — | Weights only |
 Dolphin 2.9 2 Qwen2 72B by Fireworks | Qwen2-72B-Instruct | Uncensored | — | — | Weights only |
Dolphin 2.9.1 Llama 3 70B by Cognitivecomputations | Llama 3 70B Instruct | Fine-tuned | — | — | Weights only |
Gemma 4 26B A4B It Chimeraxheretic by nanogpt | Gemma 4 26B A4B IT | Abliterated | — | — | Weights only |
Gemma 4 26B A4B Uncensored by nanogpt | — | Uncensored | — | — | Weights only |
GLM-5.3 Flash Apex Abliterated by nanogpt | GLM-5.3-Flash | Abliterated | — | — | Weights only |
 MiMo V2.6 Flash Abliterated by Xiaomi | — | Abliterated | — | — | Weights only |
MiMo V2.6 Flash Uncensored by Xiaomi | — | Fine-tuned | — | — | Weights only |
m Neural Daredevil 8B abliterated by mlabonne | — | Abliterated | — | — | Weights only |
Qwen 2.5 32B Abliterated by huihui-ai | Qwen2.5-32B-Instruct | Abliterated | — | — | Weights only |
Qwen3.5 27B Bluestar V3 Derestricted by nanogpt | Qwen3.5 27B | Uncensored | — | — | Weights only |
Qwen3.5 27B Queen Derestricted by nanogpt | Qwen3.5 27B | Uncensored | — | — | Weights only |
Frequently asked questions
A model with the refusals and content filters of the original taken out: by removing the refusal direction from its weights (abliteration), by fine-tuning on unfiltered data, or by merging models. What it knows stays the same; what it agrees to answer changes.
No. AIModelFind collects public information only: who released each model, what it is based on, where its weights are and which providers host it. The downloads and APIs belong to their authors and providers.
Most come as GGUF or MLX files, which run locally in tools such as llama.cpp, Ollama or LM Studio. Pick a quantization that fits your memory; the larger the file, the closer it stays to the full model.
Can't find what you're looking for? Contact us