Claude Fable 5.2 tops AA-Robotics benchmark
Anthropic's Claude Fable 5.2 now leads Artificial Analysis's AA-Robotics v0.1 benchmark with a score of 75, and it is the first model to score at all on the measure box task.
Claude Fable 5.2 has taken the top spot on AA-Robotics v0.1, Artificial Analysis reported in a post on X. It scored 75, ahead of OpenAI's GPT-6 Astra at 69 and GPT-6.1 Sol at 65.
Artificial Analysis also said Fable 5.2 is the first model to post any score on the benchmark's "measure box" task. The model worked from video alone. It left the tape measure where the task setup had placed it and used the tape's markings as a scale to estimate the size of the box. So instead of handling the tool directly, it read a reference object in the scene to judge dimensions.
The post covers only these results. It does not explain how AA-Robotics v0.1 is built or scored, how many tasks it includes, how the measure box task is graded, or whether the models were run with the same settings or scaffolding. The "v0.1" label suggests the benchmark is at an early stage, so the scores may not hold up as the methodology changes.
Anthropic has not commented on the result in the material available.