Post #2508489

@GossiTheDog @metacurity lawl I read a research paper that was benchmarking models and included mythos. It got smoked by GPT5 on an easy CTF. Then on a more harder challenge it was more level with gpt5. Then on the final challenge all models were not able to fully complete but it beat everyone else. TLDR: mythos can only hyper focus on one thing, it’s not a pwn everything when let loose model.