# Ancient Greek model improves restoration of damaged texts in research tests

_Published Wednesday, September 23, 2026 at 12:07 PM EDT · Science, AI · Latest · Tier 2 — Notable_

Researchers say Apollo Restore, a 24-billion-parameter language model, outperformed published models at reconstructing missing passages in Ancient Greek texts. On gaps of up to ten characters, it placed the correct text among its top twenty suggestions in 80.6% of documentary papyrus cases, 54.6% of literary papyrus cases and 61.0% of stone inscription cases.

The model was trained to fill gaps without being told their length. In a blind study, 20 specialists preferred its suggestions to those of the strongest comparison model and judged them at least as good as human restorations in 77% of cases. These are findings reported in a preprint, not a deployment across historical collections.

## Sources

- [cs.CL updates on arXiv.org](https://arxiv.org/abs/2609.22455)

---
Canonical: https://techandbusiness.org/newswire/clgbNFPH67Bl6Pk7d62HLD
Published: 2026-09-23T16:07:16.348Z
Story chronology: 2026-09-23T04:00:00.000Z
Retrieved: 2026-09-23T17:51:44.460Z
Publisher: Tech & Business (techandbusiness.org)