SciIR-Bench
Explore score history ↗SciIR-Bench evaluates text-to-image models on scientific image reasoning across three semiotic-aligned tracks: entity structure, scientific process, and scientific law. It uses an atomic checklist to convert scientific accuracy into verifiable fine-grained questions.
- Released
- 2026-06-29
- Evaluates
- General AI, Reasoning, Vision & 3D
- Openness
- unknown
- Importer
- Claire Radar
- Review status
- unreviewed
SciIR-Bench paper and code
No reported scores are on record for this benchmark yet.