VisTarget-Bench
Explore score history ↗VisTarget-Bench is a 150-task human-verified benchmark pairing each multimodal search question with a held-out target image to separate image-retrieval failures from visual-perception failures.
- Released
- 2026-08-28
- Evaluates
- General AI, Information retrieval, Multimodal, Planning
- Openness
- unknown
- Importer
- Claire Radar
- Review status
- ai-name-audit-deferred
VisTarget-Bench paper
No reported scores are on record for this benchmark yet.