GPT-OSS 120B Uncensored: Zero Refusals Reported
GPT-OSS 120B Uncensored is an open-source language model reportedly designed without content restrictions, claiming to fulfill all user requests without
Someone released an aggressively uncensored version of GPT-OSS 120B that reportedly refused zero queries during testing.
The model lives at https://huggingface.co/HauhauCS/GPTOSS-120B-Uncensored-HauhauCS-Aggressive - it’s a 117B parameter MoE (only 5.1B active) with 128K context window.
Critical settings to avoid broken outputs:
--temp 1.0 --top-k 40Disable top_p, min_p, and repeat penalties (many clients enable these by default).
llama.cpp users need this:
llama-server -m model.gguf --jinja -fa -b 2048 -ub 2048The --jinja flag is required for the Harmony response format or everything breaks.
Single 61GB file fits on one H100. Low VRAM? Use --n-cpu-moe N to offload MoE layers to CPU.
Works with llama.cpp, LM Studio, and Ollama. The creator also has smaller uncensored models (20B, 4.7B, 8B) at **https://huggingface.co/Ha
Related Tips
DeepSeek Quietly Tests Updated Model with Recent Knowledge
DeepSeek conducts quiet testing of an updated AI model that incorporates more recent knowledge and information, potentially improving its capabilities beyond
Kyutai's Hibiki Zero: 3B Speech-to-Speech Model
Kyutai introduces Hibiki Zero, a compact 3-billion-parameter speech-to-speech model that processes and generates audio directly without intermediate text
DeepSeek V4-Lite Spotted with 1M Token Context
DeepSeek V4-Lite has been observed featuring a one million token context window, significantly expanding its capability to process and analyze extremely large