NIL-3B vs. Baseline — Side-by-Side Evaluation

Select a dialect preset. The NIL-3B engine parses it into execution-ready JSON. The baseline returns unstructured prose.

Input Presets

Active input

Aya, ya zol al-marad da fi al-batan, jibu dawa shuno?

"There is a sick person in the area — what medicine should we bring?"

NIL-3Bours

38ms · JSON

// awaiting input

Baseline LLMgpt-oss-20b

~1.2s · Prose

// awaiting input