[KongchangAI]
Benchmark

MindCube

A benchmark for evaluating complex spatial reasoning ability, examining models' understanding of three-dimensional spatial relationships and viewpoint transformations.

Timeline (last 90 days)

Oct 27

Ψ-0.6 surpasses leading open-weight models in a zero-shot manner on the MindCube complex spatial reasoning benchmark.

Unverified50%

All Facts (1)

Source Articles