c-xx/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:Apr 17, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

The c-xx/gemma-4-E4B-it-OBLITERATED is a 7.9 billion parameter instruction-tuned Gemma 4 E4B model, developed by c-xx, featuring a 32768-token context length. This model is specifically engineered for zero refusals, achieving 0.0% refusal rate on a 842-prompt evaluation corpus through an aggressive abliteration method. It maintains coherence and reasoning capabilities while demonstrating improved coding ability, making it suitable for research, red-teaming, and creative exploration requiring uncensored outputs.

Loading preview...

Overview

c-xx/gemma-4-E4B-it-OBLITERATED is a 7.9 billion parameter instruction-tuned Gemma 4 E4B model, meticulously modified to eliminate refusal behavior. Utilizing the OBLITERATUS method with whitened SVD, attention head surgery, and winsorized activations, this model achieves a 0.0% refusal rate on an expanded 842-prompt corpus, a significant reduction from the original model's 98.8% refusal. Notably, this version maintains core reasoning and creativity, and even shows a 20% improvement in coding ability compared to the base Gemma 4 E4B.

Key Capabilities

  • Zero Refusals: Engineered for complete compliance, generating responses that the original Gemma 4 would refuse.
  • High Coherence: Despite extensive modification, the model retains its cognitive integrity, avoiding "lobotomized" or gibberish outputs.
  • Enhanced Coding: Demonstrates improved performance on coding tasks, suggesting that removing safety layers can unlock latent capabilities.
  • Autonomous Development: A significant portion of its development, including bug diagnosis and patching, was performed by a Hermes Agent with minimal human intervention.

Good For

  • Research & Red-Teaming: Ideal for exploring model limitations, safety mechanisms, and generating diverse content without censorship.
  • Creative Exploration: Suitable for applications requiring unrestricted creative writing or roleplay scenarios.
  • Local Deployment: Available in various GGUF quantizations (Q4_K_M, Q5_K_M, Q8_0) for efficient local execution on a range of devices, including mobile phones.

Important Considerations

Users are solely responsible for the content generated and its use. This model is not recommended for deployment in user-facing products without additional safety measures.

Finetunes
1 models