This model is based on Ground Truth from the manuscripts Cod. Vind. Slav. 3 (Breviary of Vid of Omišalj) and II. beramski brevijar. It can be used for transcribing different hands of 14th and 15th century Croatian Glagolitic handwriting. The model automatically transcribes into the Latin script and is capable of dealing with ligatures and resolving the most common abbreviations. It was trained at the Slavic Department Freiburg (Achim Rabus). The GT data was provided by Sanja Zubčić (Rijeka) and Jagoda and Guido Kappel (Vienna). About 171 000 words have been trained and the CER on the Validation Set is 5.60%.