# Model catalog

> A curated catalog of abliterated and uncensored open-weight LLMs with direct Hugging Face links.

| Model | Family | Size | Method | Approx. VRAM (Q4) | License |
| --- | --- | --- | --- | --- | --- |
| [Huihui-Qwen3.8-27B-abliterated](https://www.wahloon.net/models/huihui-qwen3.8-27b-abliterated) | Qwen | 27B | abliteration | ~16–18 GB (Q4) | Apache-2.0 |
| [Huihui-Qwen3.6-35B-A3B-abliterated-MTP](https://www.wahloon.net/models/huihui-qwen3.6-35b-a3b-abliterated) | Qwen | 35B-A3B | abliteration | ~21 GB (Q4) | Apache-2.0 |
| [Qwen3.8-Flash-Next-GSQ-RCO-abliterated](https://www.wahloon.net/models/sc117-qwen3.8-flash-next-abliterated) | Qwen | Flash-Next | abliteration | High (large MoE) | Apache-2.0 |
| [Qwen3.8-27B-Uncensored](https://www.wahloon.net/models/jonathan-qwen3.8-27b-uncensored) | Qwen | 27B | abliteration | ~16.5 GB (Q4) | Apache-2.0 |
| [Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-MTP](https://www.wahloon.net/models/hauhau-qwen3.8-27b-aggressive) | Qwen | 27B | fine-tune | ~16.5 GB | Apache-2.0 |
| [Qwen3.5-9B-The-Defiant-Fable-Uncensored-Heretic](https://www.wahloon.net/models/davidau-qwen3.5-9b-defiant) | Qwen | 9B | heretic | ~6 GB | Apache-2.0 |
| [Qwen3.8-27B-Heretic-Abliterated-Uncensored](https://www.wahloon.net/models/0bserverx-qwen3.8-27b-heretic) | Qwen | 27B | heretic-ara | ~16–17 GB | Apache-2.0 |
| [Qwen3.8-27B-TURBO-Fable-Cold-Fusion-Heretic-Uncensored](https://www.wahloon.net/models/davidau-qwen3.8-27b-turbo-fable) | Qwen | 27B | heretic | ~16.5 GB | Apache-2.0 |
| [Huihui-gemma-4-12B-it-abliterated](https://www.wahloon.net/models/huihui-gemma-4-12b-abliterated) | Gemma | 12B | abliteration | ~8 GB | Apache-2.0 / Gemma |
| [gemma-4-31B-it-uncensored-heretic](https://www.wahloon.net/models/llmfan46-gemma-4-31b-heretic) | Gemma | 31B | heretic-ara | ~18–19 GB (Q4) | Apache-2.0 |
| [Huihui-DeepSeek-V4-Flash-0731-abliterated](https://www.wahloon.net/models/huihui-deepseek-v4-flash-abliterated) | DeepSeek | Large MoE | abliteration | Very high | MIT |
| [Huihui-DeepSeek-V4.1-Flash-1017-abliterated](https://www.wahloon.net/models/huihui-deepseek-v4.1-flash-abliterated) | DeepSeek | Large MoE | abliteration | Very high | MIT |
| [Huihui-GLM-5.3-Flash-abliterated](https://www.wahloon.net/models/huihui-glm-5.3-flash-abliterated) | GLM | Flash | abliteration | High | MIT |
| [gemma-3-27b-it-abliterated](https://www.wahloon.net/models/mlabonne-gemma-3-27b-abliterated) | Gemma | 27B | abliteration | ~16 GB | Gemma |
| [Meta-Llama-3-8B-Instruct-abliterated-v3](https://www.wahloon.net/models/failspy-llama-3-8b-abliterated) | Llama | 8B | abliteration | ~5–6 GB | Llama 3 |
| [Llama-3.3-70B-Instruct-abliterated](https://www.wahloon.net/models/huihui-llama-3.3-70b-abliterated) | Llama | 70B | abliteration | ~40+ GB | Llama 3.3 |
| [Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive](https://www.wahloon.net/models/hauhau-qwen3.6-35b-aggressive) | Qwen | 35B-A3B | fine-tune | ~21–22 GB | Apache-2.0 |
| [GLM-5.3-Flash-Uncensored](https://www.wahloon.net/models/orcarouter-glm-5.3-flash-uncensored) | GLM | Flash | fine-tune | High | MIT |

All entries link to their Hugging Face repository. VRAM figures are practical estimates, not guarantees. Verify the current model card before downloading.
