[KongchangAI]
Benchmarkpelican riding a bicycle / 鹈鹕骑自行车

Pelican Riding a Bicycle

An informal benchmark used by Simon Willison since 2024 to compare the drawing abilities of large models. It asks a model to generate an SVG vector image of a pelican riding a bicycle, testing the model's ability to understand and combine animal anatomy, mechanical structures and spatial relationships.

Timeline (last 90 days)

Oct 22

In communities around AI-generated content, "a pelican riding a bicycle" has become an informal test case for probing generative video models' ability to understand absurd combinations.

Unverified50%
Oct 22

The "pelican riding a bicycle" test can simultaneously examine a generative video model's compositional generalization, object interaction understanding, and temporal consistency.

Unverified50%

All Facts (2)

Source Articles