BhashaBench-Multi is a large-scale multilingual multiple-choice question answering benchmark covering four specialised Indian knowledge domains across 22 Indian languages plus English. It was created by bharatgenai and last updated on 2026-06-09. The dataset is designed to evaluate language understanding and domain knowledge.
Use Cases
- Benchmarking multilingual language models based on coverage of 22 Indian languages
- Evaluating domain-specific knowledge in AI systems based on the four specialised Indian knowledge domains
- Training models for multiple-choice question answering tasks in low-resource languages
Strengths
- Covers 22 Indian languages plus English
- Focuses on four specialised Indian knowledge domains
- Designed as a purpose-built benchmark for Indian domain knowledge
Limitations
- Column-level documentation is absent; field semantics must be inferred after download
- Row count is unknown, which may limit suitability assessment
Provenance
- Source
- bharatgenai
- Freshness
- Last updated 2026-06-09 12:45:35
- Geography
- India