12,125 queries for benchmarking AI tool-use in Arabic, covering five dialects and eight real-world domains. The dataset was created by TuwaiqAcademy and a test set is scheduled for release on July 20, 2026.
Use Cases
- Benchmarking Arabic AI agent performance based on queries across five dialects
- Training function-calling models based on 27 structured tools
- Evaluating cross-dialectal AI comprehension based on eight real-world domains
Strengths
- 12,125 queries provide a substantial benchmark size
- Covers five Arabic dialects and eight distinct domains
- Includes 27 structured tools for function calling
Limitations
- Description metadata is limited; actual data quality requires manual inspection after download
- Column-level documentation is absent; field semantics must be inferred after download
Provenance
- Source
- TuwaiqAcademy
- Freshness
- Last updated 2026-05-16 09:25:48; freshness should be verified
- Geography
- Covers Arabic dialects from regions including Cairo and Riyadh