AISA-ArabicFC
Arabic Function Calling for Agentic AI Systems
The first open benchmark for tool-use in Arabic — across five dialects, eight real-world domains, and 27 structured tools.
12,125 queries · 5 dialects · 8 domains · 27 tools · 12K reasoning traces
📅 Test set releases July 20, 2026 · 🏛️ Budapest · Oct 24–29, 2026
Argument scoring is now robust to surface form. A correct… See the full description on the dataset page:
https://huggingface.co/datasets/TuwaiqAcademy/AISA-ArabicFC.
