Fully decensored Gemma 4 12B (abliterated with Heretic) — 0/100 genuine refusals at KL 0.0284, i.e. near-zero capability loss.
3,189 Pulls 3 Tags Updated 2 days ago
Decensored (abliterated) Gemma 4 12B — refusals removed with Heretic while keeping the base model's intelligence (KL 0.0154, 0 true refusals). For uncensored assistant and 18+ companion/roleplay use.
317 Pulls 4 Tags Updated 17 hours ago
Quantized GGUF of hf.co/OpenYourMind/gemma-4-12B-it-abliterated-uncensored
198 Pulls 3 Tags Updated 17 hours ago
Quantized GGUF of hf.co/DuoNeural/Gemma4-12B-IT-Abliterated
29 Pulls 3 Tags Updated 17 hours ago
Cipher-Abliterated is an adaptive conversational AI system designed for flexible interaction, enhanced creative reasoning, and less restrictive conversational behavior while maintaining structured output and technical usability. It is optimized for users
277.5K Pulls 1 Tag Updated 1 week ago
1,917 Pulls 1 Tag Updated 2 weeks ago
614 Pulls 1 Tag Updated 2 weeks ago
407 Pulls 1 Tag Updated 1 week ago
443 Pulls 1 Tag Updated 2 weeks ago
NullGuard is an abliterated research build from gemma4:e4b by h4rithd, made for academic study of local LLM customization, refusal behavior, and safety limits. For research, education, and controlled security testing only.
401 Pulls 1 Tag Updated 3 weeks ago
210 Pulls 1 Tag Updated 1 week ago
Gemma 4 models are designed to deliver frontier-level performance at each size. They are well-suited for reasoning, agentic workflows, coding, and multimodal understanding.
244.8K Pulls 29 Tags Updated 1 month ago
198 Pulls 1 Tag Updated 2 weeks ago
Qwen 3.5 is a family of open-source multimodal models that delivers exceptional utility and performance.
235.2K Pulls 58 Tags Updated 2 months ago
Qwen3.6 delivers substantial upgrades in agentic coding and thinking preservation than previous Qwen models.
76.8K Pulls 17 Tags Updated 1 month ago
75 Pulls 1 Tag Updated 1 week ago
As the strongest model in the 30B class, GLM-4.7-Flash offers a new option for lightweight deployment that balances performance and efficiency.
55.7K Pulls 5 Tags Updated 4 months ago
Abliterated (decensored) version of google/gemma-4-E4B-it, created using Heretic v1.2.0 with Arbitrary-Rank Ablation (ARA) and row-norm preservation.
9,962 Pulls 5 Tags Updated 2 months ago
Qwen3.6-35B-A3B-Claude-4.7-Opus-abliterated-fulluncen
7,190 Pulls 2 Tags Updated 1 month ago
12.5K Pulls 1 Tag Updated 3 months ago