(TibetanReview.net, Nov05’25) – The Monlam Tibetan IT Research Centre of exile Tibetans based in India has said at its third annual event on Nov 3 in Dharamshala, India, that its Melong.ai surpasses existing large language models such as Gemini, Claude and DeepSeek in the Tibetan‑language generation and comprehension.
While launching four major digital tools at its 2025 Monlam Manifestation event held at the Tibetan Institute of Performing Arts (TIPA), the centre has said its newly upgraded “Monlam Think”, an updated version of Monlam Melong, is particularly noteworthy for surpassing global AI models in Tibetan language comprehension and analysis.
In internal testing, the updated model reportedly answered about 1,500 questions on Tibetan history, literature and grammar, achieving a score of 73%. Beside, it scored 71% on the CBSE Class 12 Tibetan exam, outperforming other AI models in both cases, Tenzin Gayche, AI instructor at the Monlam IT centre, has said.
“Previously, Monlam Melong could only process a single paper at a time. Now, it can ingest an entire book for summarisation and analysis,” covering Tibetan history, literature, and grammar, Gayche has added.
Melong.ai is also multilingual and multimodal, supporting more than 140 languages and accepting both image and text input.
Monlam Melong is part of Monlam IT team’s pioneering AI work, aimed not just at making Tibetan digitally accessible, but also at preserving the language and the knowledge it carries in a way that reflects Tibetan perspectives and understanding.
And so, Monlam IT has also developed several other initiatives to support Tibetan language and literature digitally. Monlam Bodyig-4 now features three additional fonts and improved typesetting, making tantric texts clearer and more accessible to readers. The Monlam Grand Dictionary serves as a comprehensive reference for Tibetan language learners and scholars, providing detailed explanations and usage examples.
Monlam AI also offers intelligent support for research, language learning, and text analysis, powered by machine-learning models for translation, OCR, speech-to-text, and text-to-speech. It enables users to engage with Tibetan content more efficiently and interactively and it now supports 9 languages, expanding its accessibility to a wider audience.
Geshe Lobsang Monlam, the founder-CEO of Monlam Tibetan IT Research Centre, has outlined his vision for the next decade, saying the Centre aims to rely on the power of Monlam AI from 2025 to 2035 to integrate Tibetan religion, culture, and modern knowledge into the world’s technological progression.