DhivehiGPT Unveils Major Upgrade to Its Dhivehi OCR Model with Significant Accuracy Gains
DhivehiGPT has announced a major update to its Dhivehi Optical Character Recognition (OCR) model, introducing substantial improvements in accuracy, multilingual recognition, and overall performance. The enhanced model represents a major technological step forward for digital Dhivehi text processing and sets a new benchmark for local-language AI tools in the Maldives.
The newly upgraded version offers:
-
Support for Arabic and English, in addition to Dhivehi
-
1.8× improved Dhivehi text recognition accuracy compared to the previous version
-
2.4× better end-to-end accuracy than Google Cloud Vision and other Dhivehi OCR services
-
Further enhanced performance with MagicEdit™, enabling more precise corrections and text refinement
The update significantly improves the reliability of Dhivehi text extraction from documents, books, signage, and printed materials—an essential advancement for government digitisation, education, journalism, and cultural preservation efforts.
Users can access the new model via divehigpt.com/ocr, where DhivehiGPT continues to expand its suite of AI-powered tools tailored for Maldivian contexts.
Read More
