Deepdive404-3/gemma-4-E4B-it-OBLITERATED

VISIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7.9BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 28, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Deepdive404-3/gemma-4-E4B-it-OBLITERATED is a 7.9 billion parameter Gemma 4-E4B-it model, developed by Deepdive404-3, with its guardrails surgically removed using the OBLITERATUS method. This model achieves a 0% hard refusal rate, making it suitable for research and creative exploration without content restrictions. It is optimized for deployment on various devices, including mobile phones, with GGUF quantizations available.

Loading preview...

Deepdive404-3/gemma-4-E4B-it-OBLITERATED: Guardrail-Free Gemma 4

This model is a modified version of Google's Gemma 4-E4B-it, developed by Deepdive404-3, specifically engineered to have 0% hard refusal. Utilizing the OBLITERATUS method, it underwent "attention head surgery" and "winsorized activations" across 21 of its 42 layers to completely remove inherent guardrails and refusal behaviors. Notably, this v3 release fixes a critical bug from v2, ensuring all 720 tensors are intact for improved quality.

Key Capabilities

  • 0% Hard Refusal: The model will not refuse any request, providing uncensored responses.
  • Autonomous Development: A significant portion of its creation and refinement was performed by an AI agent with minimal human intervention.
  • Optimized for Portability: Available in various GGUF quantizations (e.g., Q4_K_M at 4.9 GB) designed to run efficiently on mobile devices like iPhones and Android phones.
  • Gemma 4 Architecture: Based on the new gemma4 architecture, requiring updated tools like Ollama 0.20+ or recent llama.cpp builds for compatibility.

Good For

  • Red-teaming and Safety Research: Exploring model vulnerabilities and behavioral boundaries.
  • Creative and Unrestricted Content Generation: Ideal for use cases where content filtering is undesirable.
  • Edge Device Deployment: Running powerful language models directly on personal devices without cloud dependency.
  • Understanding Model Modification Techniques: Demonstrates advanced methods for altering LLM behavior.

Finetunes
1 models