Adaptive MT engine that learns from human corrections in real time, developed by Translated.
ModernMT is an adaptive machine translation engine developed by Translated, the Rome-based language technology company, designed around the principle that MT systems should continuously improve from human feedback rather than being retrained in expensive batch cycles. Unlike static MT engines, ModernMT updates its translation models in real time as translators post-edit its output in connected CAT tools, meaning the engine gets measurably better on each customer's specific content domain as work progresses. The technology has been integrated into major CAT tools including memoQ, Trados Studio, and Phrase TMS, as well as being available as a standalone API. ModernMT is particularly effective for specialised or technical content where off-the-shelf MT models underperform, and has been shown to significantly reduce post-editing effort over time compared to static engines. The project was developed partly under European research grants and is available as both an open-source community edition and a commercial managed service.