Users say Mythos/Fable 5.2 stands out on multi-task performance; @claudeai says Fable 5.2 leads on RoboDojo
X user @logangraham said that after using Mythos/Fable 5.2 internally, they believe the model stands out on writing, 3D modeling, video production, one-shot interactive tools and robotics tasks, and said it is best suited for cyber defense. The cited @claudeai post said Fable 5.2 shows significantly improved skill at controlling physical robots, even on tasks it was not explicitly trained on; on the RoboDojo benchmark, which simulates a two-armed robot, it outperformed other models and most AI systems built specifically to control robots.
from using Mythos/Fable 5.2 internally, this is one of those “big model smell” models — has a je ne sais quoi at writing, 3d modeling, making videos, 1-shotting interactive tools, and robotics. Also the best for cyberdefense.
Fable 5.2 shows substantially improved skills in controlling physical robots, even on tasks it has never explicitly been trained to do.
On RoboDojo, a benchmark with a simulated two-arm robot, it outperformed other models and most AI systems built to control robots.View the quoted post on X