OpenMythos benchmarks are now available, evaluating performance on SWE-bench Pro, CyberGym, and cybench. The results show the model performs well for a small cybersecurity-focused model, though further training is planned to improve capabilities. GGUF versions and demo links are provided on Hugging Face.