The Austrian Academy of Sciences (OeAW) has launched a bold initiative to democratize access to ancient history through artificial intelligence. In collaboration with Mistral AI and Sail Reply, a Reply Group Company, the academy is developing Apollo—the world's first advanced multimodal Large Language Model designed specifically for Ancient Greek. Named after the Greek god of light and patron of the arts and sciences, Apollo aims to transform how researchers interact with thousands of years of fragmented human knowledge.

The Scale of Unread History

The project addresses a staggering problem in classical scholarship: there are approximately one million Greek papyri worldwide that have never been read. Tens of thousands of these irreplaceable historical artifacts sit in the Papyrus Collection of the Austrian National Library alone, their contents inaccessible to modern researchers without years of specialized expertise. Apollo is being trained on roughly 600 million words from historical Greek texts and tens of thousands of published inscriptions and papyri—a dataset that represents the largest digital corpus of historical Greek assembled to date.

Technical Architecture and European Infrastructure

Unlike conventional LLMs built for modern languages, Apollo must contend with a historical language that evolved significantly across centuries. The model is being developed on secure European infrastructure provided by Sail Reply, ensuring compliance with strict data protection requirements while leveraging Mistral AI's generative capabilities. This technical foundation enables researchers to perform advanced semantic and thematic searches across the entire Greek textual tradition—a task previously requiring decades of specialized study.

Pioneering Work in Historical Language Processing

Anna Dolganov, an ancient historian and papyrologist at the Austrian Archaeological Institute of the OeAW, is leading the project. Her role involves providing field-specific guidance, overseeing integration of relevant sources, and guaranteeing scientific quality through rigorous historical contextualization. "This AI system can be developed in many directions for a wide range of research tasks," Dolganov explained, "from reconstructing fragmentary inscriptions and papyri to conducting semantic and thematic searches across the entire Greek textual tradition to deciphering handwritten texts."

Accelerating Decades of Scholarship

The technology promises to compress what once took years into mere hours. Automatic text restoration capabilities allow Apollo to fill gaps in heavily damaged documents, while advanced searching enables researchers to identify connections across disparate sources instantly. According to OeAW President Heinz Faßmann: "Ancient languages and artificial intelligence are not a contradiction. The collaboration between our Archaeological Institute and Mistral and Reply demonstrates how AI is advancing research in the Humanities."

Key Takeaways

  • Apollo represents the first LLM developed for a historical language evolving over many centuries
  • Trained on ca. 600 million words from historical Greek texts plus tens of thousands of inscriptions
  • Secure European infrastructure through Sail Reply with Mistral AI's generative models
  • Project targets one million unread papyri worldwide, including significant holdings at the Austrian National Library

The Bottom Line

This is exactly the kind of cross-domain innovation that demonstrates what happens when cutting-edge AI meets real-world scholarly problems. Apollo doesn't just translate Greek—it reconstructs fragmented history itself. If successful, this model sets a blueprint for similar projects across Akkadian, Latin, Sanskrit, and every other dead language drowning in unreadable manuscripts.