Insanity. Why is it so much better than Fable and GPT-5.6 Sol in Frontier-Bench? The ARC-AGI-3 score is insane too, wtf. And simultaneously kinda stagnating on DeepSWE.
This all smells like a completely different (new) base model rather than a Fable/Mythos distillation to me. Don't know how this could make sense otherwise
Insanity. Why is it so much better than Fable and GPT-5.6 Sol in Frontier-Bench? The ARC-AGI-3 score is insane too, wtf. And simultaneously kinda stagnating on DeepSWE.
This all smells like a completely different (new) base model rather than a Fable/Mythos distillation to me. Don't know how this could make sense otherwise