🐉 Qwythos-9B Claude Mythos Chat (ZeroGPU)

Powered by empero-ai/Qwythos-9B-Claude-Mythos-5-1M — a 9B uncensored reasoning model trained on Claude Mythos/Fable traces.

• 1M context (YaRN) · Native function calling · Strong technical / CoT reasoning
• GGUF quants for local use: empero-ai/Qwythos-9B-Claude-Mythos-5-1M-GGUF
• Running on Hugging Face ZeroGPU (free on-demand GPU). Quota applies.

The model usually writes a <think>...</think> reasoning block before the final answer.

Tips

• Lower temperature (0.3–0.5) for more deterministic technical answers; 0.6 is the model default.
• Increase Max New Tokens if the model is still inside a long <think> block.
• For local offline use, download the GGUF files and run with llama.cpp / Ollama / LM Studio.
• ZeroGPU quota resets daily. PRO users get higher limits + can buy credits.