The Austrian Academy of Science is releasing Apollo, the world's first advanced artificial intelligence language model for Ancient Greek, developed in partnership with French AI lab Mistral and Sail Reply. Trained on approximately 600 million historical Greek words from manuscripts, papyri, and inscriptions, Apollo is designed to help scholars restore fragmented ancient texts by intelligently filling in missing or illegible passages. This represents a significant acceleration of work that previously required rare specialist expertise and has been described as transformative for academic research.

The model works by recognising linguistic patterns and contexts—adapting to different dialects and styles such as Homeric or Doric Greek—and proposing multiple word options for scholars to select between, ensuring human expertise remains central to the process. Previously, papyrus restoration demanded skilled academics to identify word divisions, accurately date documents, weigh socio-political contexts, and consult reference materials. Whilst the technology is unlikely to dramatically alter broad historical understanding, as many unrestored papyri contain mundane materials such as personal letters and contracts, it could uncover valuable details about daily life in antiquity and substantiate existing scholarly assumptions. The developers emphasise that human competence must remain paramount to prevent the historical record from being polluted with AI errors, and the model is built with this safeguard in place.