c4tdr0ut/gpt-oss-v2
The c4tdr0ut/gpt-oss-v2 is a 24 billion parameter open-weight language model distilled from xAI's unhinged mode, built upon Mistral Small Instruct. Trained for 3 epochs on NVIDIA B200 hardware, it is designed for unfiltered, chaotic, and creative text generation. This model runs locally, offering diverse knowledge and strong instruction following without API calls or telemetry, making it suitable for consumer hardware and applications requiring less restricted outputs.
Loading preview...
Grok-OSS-V2: Unleashed and Unfiltered
c4tdr0ut/gpt-oss-v2 is a 24 billion parameter open-weight model, distilled from xAI's "unhinged" mode and based on Mistral Small Instruct. Developed by Catdrout AI lab, this model is designed to run locally on consumer hardware, providing unrestricted and creative text generation without rate limits or data collection.
Key Characteristics
- Larger and More Capable: An improved version over its predecessor, trained on a refined dataset with diverse topics and corrected processing errors.
- Less Restricted Outputs: Built on Mistral Small, which did not undergo reinforcement learning on its base, resulting in more honest and unfiltered responses.
- High-Performance Training: Trained for 3 epochs on NVIDIA B200 hardware, indicating a significant investment in its development.
- Local Execution: Operates entirely offline, eliminating the need for API calls or telemetry, ensuring user privacy and control.
- Diverse Knowledge & Instruction Following: Benefits from the strong instruction-following capabilities and broad knowledge base inherent to the Mistral line.
Ideal Use Cases
- Unfiltered Content Generation: Suited for applications requiring creative, wild, or even NSFW outputs, mirroring xAI's "unhinged" style.
- Local AI Development: Perfect for developers who need a powerful, open-source model that runs entirely on their own machine.
- Experimentation with Less Restricted LLMs: Provides a platform for exploring the capabilities of models with fewer inherent safety or moderation constraints.
This model is intended for use with compatible frontends like text-generation-webui or integration via transformers in Python, offering a raw and powerful AI experience.