StemBind Benchmark

What is StemBind?

StemBind is a diagnostic benchmark with shared-stem questions to attribute failures in abstract visual reasoning. It includes perception, rule, and full tasks…

Released

2026-05-29

Evaluates

General AI, Reasoning, Vision & 3D

Openness

unknown

Importer

Claire Radar

Review status

unreviewed

Reported scores

0

Source provenance

• Original evidence https://arxiv.org/abs/2606.00148

StemBind paper

• Paper https://arxiv.org/abs/2606.00148
• Website https://hexixiang.github.io/StemBind

Which sources cite StemBind?

• StemBind

Related General AI, Reasoning, Vision & 3D benchmarks

• $\tau^\tau$-Bench: An Environment for End-To-End, Realistic Agent Construction
• 3DCodeBench
• 5-Dialects-BN
• A-OKVQA
• A3S-Bench
• AARRI-Bench
• ActBench
• ActiveFly-Bench