Benchmark Radar™
Star —

ZeroBench

Explore score history ↗

Challenging visual reasoning benchmark designed to be impossible for frontier models at release; evaluates complex reasoning over images.

Released
2025-02-13
Evaluates
multimodal
Openness
unknown
Importer
Model reports

ZeroBench code

ZeroBench results and reported scores

Reported benchmark scores · Model reports

Reported scores: 1

ModelOrganizationReported valueReportedEvidence
Seed2.0 ProByteDance12.02026-02-14evidence

Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.

Which sources cite ZeroBench?

Related benchmarks