Model overview
What this model is for
Qwen3.8-27B-Uncensored-Cyber-NVFP4 is a 27-billion-parameter Hugging Face model intended mainly for cybersecurity-related use. It is presented as an NVFP4 variant of the Qwen3.8-27B-Uncensored-Cyber model and is associated with safetensors, compressed-tensors, vLLM, Qwen3/Qwen3.5 compatibility, and an 8-bit tag. These labels suggest a deployment-oriented checkpoint, particularly for users evaluating compressed or quantized inference workflows, but the supplied source does not provide setup instructions, tested hardware, context-window information, or inference benchmarks. The model is explicitly labeled “Uncensored,” so developers should independently evaluate output behavior, safeguards, and suitability for their application rather than assume standard content controls. No modalities are disclosed, so image, audio, and video capabilities should not be assumed. Pricing, token throughput, and detailed cybersecurity task results are not disclosed. The listed license is Apache-2.0, although developers should also verify the licensing and usage terms of any base model or included components before redistribution or commercial deployment. The source is the model’s Hugging Face page; its availability is confirmed, but the supplied page extract contains limited model-card detail and does not establish a publication or update date.