Data as of Jul 25, 2026 · Based on 2,718,867 AI responses across 9,511 prompts · See how Parse measures this
BESSTIEgithub.com · Low evidenceBESSTIE is a benchmark dataset for sentiment and sarcasm classification across Australian, Indian, and British English varieties. It includes manually annotated Google Place reviews and Reddit comments, and evaluates nine large language models to reveal performance gaps between inner-circle and outer-circle English varieties.
Parse Score