ProcessBench-2026other

ProcessBench

A process-aware VLM benchmark for robotic manipulation containing 57,892 QA items across 12 task families for evaluating vision-language model understanding of manipulation processes.

Downloads92

Technical Profile

Modalities
rgblanguage
Environment
lab
Task Types
manipulation
Data Format
parquet
License
other
Part of the ProcessBench family

Access

Need custom rgb data?

Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.

Request a Sample Pack

Related Datasets