Anthropic says Fable 5.2 leads on RoboDojo robot-control benchmark

Anthropic says its Fable 5.2 model outperforms other models and most robot-control AI systems on RoboDojo, a benchmark using a simulated two-arm robot.

The company also says the model shows improved robot control, including on tasks it was not explicitly trained for. Alex Albert, who is affiliated with Anthropic, says robotics will follow the bitter lesson, the view that general methods scaled with compute eventually beat hand-built approaches, and cites the Fable 5.2 results in an X post.

The robot-control claims come from a quoted @claudeai post. The reports include no independent evaluation and no further benchmark detail. Separately, Anthropic says Fable 5.2 and Mythos 5.2 are a step above Opus 5.5 in almost all capabilities it evaluated, and that they generate outputs up to 30% faster than Fable 5.1 and Mythos 5.1.