uncensored open-source LLM built for pentesting
Posted by BreakfastAdept9758@reddit | LocalLLaMA | View on Reddit | 8 comments
WhiteRabbitNeo was created in 2023 for offensive and defensive cybersecurity — no guardrails, so it actually helps with things mainstream models refuse to touch. It got acquired by Kindo in 2024 and rebranded as Deep Hat, but the original models are still publicly available.
Open-source models: https://huggingface.co/WhiteRabbitNeo
llama-impersonator@reddit
use an abliterated model, these old models are terrible
BreakfastAdept9758@reddit (OP)
What abliterated models would you recommend for pentesting specifically? Is Qwen3.5 good for security tasks or more general purpose?
Additional-Curve4212@reddit
Hey guys is there a good uncensored llm with the main goal of making malicious/phishing stuff?
Ok_Mammoth589@reddit
These models sucked then and I'm certain they suck now. It still looks like they're still just trying to float on a rising tide, fine tuning the next model release.
shikima@reddit
I use qwen3.5 9b, works fine there are 3-4 uncensored ones
Purple-Programmer-7@reddit
Thank you for giving us open weights Mythos
Miriel_z@reddit
Added to check, thanks! I am searching for a good coding model at around 8B-12B. Any pointers?
Purple-Programmer-7@reddit
Gonna be tough. Look at the new Gemma 4 models.