Translated Announces API Access to Lara Through TranslationOS
Translated integrates Lara AI into TranslationOS, enabling accurate, brand-consistent translations in 200+ languages with real-time API support.
→ Continue ReadingL
ara surpasses traditional machine translation (MT) by blending accuracy with the fluency, reasoning, and contextual understanding of large language models (LLMs). It delivers quality approaching that of the best professional translators, explaining its translation choices to ensure clarity and alignment with the user’s intent.
Lara represents over a decade of research in machine translation with a clear mission: to help people understand and be understood in their native language effortlessly. Lara surpasses your standard machine translation, delivering quality that approaches the level of top-tier professional translators. This is possible because Lara draws from a unique dataset of 25 million real-world translations crafted by experienced professionals. These translations were produced by MT systems and revised by professional translators, who documented errors, noted corrective actions, and shared insights when disagreements arose. Lara’s architecture includes a critique model that relies on a chain-of-thought approach to gather insights, weigh options, and revise the translation throughout the process to ensure that the original meaning is accurately conveyed in the target language.
In addition to its unique reasoning abilities, Lara stands out for its contextual understanding. Instead of translating sentence by sentence, Lara intelligently considers the entire document to ensure a natural flow and consistency. This approach captures the original intent while making the translation readable and resonant for the audience.
Additionally, Lara offers users three distinctive styles: Faithful (ideal for technical, legal, and instructional content), Fluid (for general content), and Creative (for marketing or literature). If any ambiguity arises, Lara can prompt the user for specifics, such as gender references or contextual details, to fine-tune the translation further. When necessary, Lara even provides access to language experts for an added layer of precision.
For enterprises, Lara integrates smoothly into localization workflows through an API and connects with various content sources via our platform, TranslationOS. This integration enhances localization quality, reduces costs, and significantly increases content translation capacity, enabling companies to deliver a high-quality, seamless experience to a wider global audience.
Lara is also a versatile online tool that professionals can depend on for both personal and business translations. It can quickly translate any document, even complex formats like InDesign files, in minutes. This capability is especially valuable for creatives, as it allows them to receive translated, layout ready documents without needing to start from scratch—freeing up more time for creative work.
Marco Trombetti is a computer scientist, entrepreneur, and investor. In 1999, he co-founded Translated, which pioneered the application of artificial intelligence in the language industry. Through Translated, Trombetti led the introduction of adaptive machine translation technology and fostered a symbiotic relationship between professional translators and AI. Leveraging the success of Translated and other ventures, Marco co-founded Pi Campus, a venture capital firm investing in AI-applied startups, and Pi School, an educational institution dedicated to training the next generation of AI specialists.
Related Articles
Translated integrates Lara AI into TranslationOS, enabling accurate, brand-consistent translations in 200+ languages with real-time API support.
→ Continue ReadingTranslated announced a new MT model called Hyper Adaptive, which enables companies to translate billions of words at ultra-fast speeds without compromising quality.
→ Continue ReadingThe report signals a fundamental paradigm shift in AI: the transition from models trained on static, archived human knowledge to systems that learn through direct…
→ Continue Reading