The adversarial screening questions only retain samples of difficulty.

OpenAI launches IndQA benchmark: a "Context and Reasoning" evaluation set for Indian languages and cultures.

On November 3, 2025, OpenAI released IndQA, a new benchmark for evaluating AI systems' understanding and reasoning in Indian languages and cultures. T...