Ultralytics says Claude Fable 5.2 refused every baggage X-ray in its benchmark

Ultralytics says Claude Fable 5.2 refused all baggage X-ray images in its benchmark, including a scan of an ordinary phone.

The company reports that every open-weights and Chinese model it tested answered all 10 prompts. The report comes from an X post by Ultralytics, and the benchmark's methods and image set are not described in the source.