V3N0M/Aisha-Uncensored-Gemma-4

VISIONConcurrency Cost:1Model Size:5.1BQuant:BF16Ctx Length:32kTool Calling:SupportedPublished:Apr 7, 2026Architecture:Transformer0.0K Cold

Aisha-Uncensored-Gemma-4 is a 5.1 billion parameter Gemma-based model developed by V3N0M, fine-tuned and converted to GGUF format using Unsloth. This model is designed for uncensored text generation and includes multimodal capabilities, supporting both text-only and vision-enabled applications. It offers various quantized GGUF files for flexible deployment, making it suitable for local inference with diverse hardware configurations.

Loading preview...

Model Overview

Aisha-Uncensored-Gemma-4 is a 5.1 billion parameter model based on the Gemma architecture, developed by V3N0M. It has been fine-tuned and optimized using Unsloth, which facilitated a 2x faster training process. This model is provided in GGUF format, making it compatible with tools like llama-cli for efficient local inference.

Key Capabilities

  • Uncensored Text Generation: Designed to provide unrestricted text outputs.
  • Multimodal Support: Includes variants that support vision inputs, indicated by mmproj files, allowing for processing of both text and image data.
  • GGUF Format: Available in multiple quantized GGUF files (e.g., F16, Q8_0, Q4_K_M, BF16-mmproj) to accommodate different performance and precision requirements.

Deployment Notes

  • Ollama Compatibility: For vision models, Ollama currently requires a unified bf16 model. Users need to create a Modelfile and use ollama create to combine the model and mmproj components into a single file for use within the Ollama ecosystem.
  • Ease of Use: Example command-line usage is provided for both text-only and multimodal inference with llama-cli and llama-mtmd-cli respectively, leveraging Jinja templating.

Finetunes
2 models