Huihui Qwythos 9B Claude Mythos 5-1M Abliterated
huihui-ai/Huihui-Qwythos-9B-Claude-Mythos-5-1M-abliterated-GGUF
published Jun 2026 · updated Jun 2026
Huihui Qwythos 9B Claude Mythos 5-1M Abliterated is a text-generation model that is an uncensored version of Qwythos-9B-Claude-Mythos-5-1M, created via abliteration, producing output without safety filtering and with strict adherence to the first prompt's instructions.
specs
| Task | Text Generation |
| Parameters | 9 billion |
| Context Length | 262,144 tokens |
| Uncensored | Yes (abliterated) |
about this model
Huihui-Qwythos-9B-Claude-Mythos-5-1M-abliterated-GGUF is a text-generation model that provides an uncensored variant of empero-ai/Qwythos-9B-Claude-Mythos-5-1M, created using abliteration to remove safety filters.
Capabilities
In non-thinking mode, the model strictly adheres to the first prompt's instructions, delivering only the requested output—whether a translation or a specific format—without any introductory or concluding fluff. This behavior makes it suitable for tasks where precise format compliance is needed.
Key Strengths
- Uncensored: safety filtering has been significantly reduced, enabling generation of content that standard models would refuse.
- Strict instruction following: outputs exactly what is requested, no extra commentary.
Usage Considerations
The model’s reduced safety filtering increases the risk of sensitive, controversial, or inappropriate outputs. It is not suitable for all audiences or public-facing production applications without thorough review. Users should monitor outputs in real-time and ensure compliance with local laws and ethical standards. This model is recommended for research, testing, or controlled environments.
Gigarouter hosts this model as a managed, OpenAI-compatible API, eliminating the need for local installation or configuration.
best for
- ·Translation tasks where exact output formatting is required
- ·Generating structured data (JSON, CSV) without extra commentary
- ·Unfiltered creative writing or role-playing in controlled research environments
FAQ
Abliteration removes safety training, resulting in an uncensored model that may produce sensitive or controversial content. Use with caution.
The model supports up to 262,144 tokens of context when using the recommended llama.cpp settings.
It is the same base model but with safety filtering removed (abliterated). It also strictly follows the first prompt, outputting only what is requested without extra fluff.
This model can generate sensitive, controversial, or inappropriate content. It is not suitable for all audiences and should be used in research or controlled environments only.
Use gigarouter's OpenAI-compatible endpoint with your API key. The model is hosted and ready for inference without local setup.
We're benchmarking and onboarding Huihui Qwythos 9B Claude Mythos 5-1M Abliterated as a hosted, OpenAI-compatible API. Sign in for free credit and be ready when it lands, or tell us you want it and we'll prioritize it.